Notes on Scandal: Lobbying Reforms and Public Opinion in Canada

A Dissertation

Presented to the

The Faculty of the Graduate School

At the University of Missouri-Columbia

______

In Partial Fulfillment

Of the Requirements for the Degree

Doctor of Philosophy

______

By

Gwendolyn E. Sams

Dr. James W. Endersby, Dissertation Supervisor

May 2016

The undersigned, appointed by the dean of the Graduate School, have examined the dissertation entitled

NOTES ON SCANDAL: LOBBYING REFORMS AND PUBLIC OPINION IN CANADA presented by Gwendolyn E. Sams,

a candidate for the degree of doctor of philosophy, and hereby certify that, in their opinion, it is worthy of acceptance.

______

Professor James W. Endersby

______

Professor Lael R. Keiser

______

Professor Laron K. Williams

______

Professor Mary Stegmaier

ACKNOWLEDGEMENTS

First and foremost, I’d like to thank my committee and especially James Endersby for remaining supportive throughout this process and encouraging me to finish. Without the help of these fine individuals, I would have never had the passion to choose such a topic and finish.

James, in particular, deserves special appreciation for his constant enthusiasm about Canada without which I never would have seen the light. For all of the mentorship, helpful criticism, and dedication to doing what you all do, I am so thankful. To the Kinder Institute, thank you for your generous funding without which this dissertation would not have been possible. I would also like to say thank you to the University of Ottawa’s Centre on Governance which hosted me and provided generous resources for my research while in Canada.

To my husband, I extend thanks for his constant support of my dreams and ideas and also for his help editing this draft. Without him and his constant distaste for adverbs, I would never have kept my sanity and could not have juggled a job and this dissertation. At the same time, this dissertation finally helped him his dreams of being a copy editor. Still, for all the extra dishes washed, dog walk sessions, and time spent painfully checking citations, this one is for you,

Daniel.

I thank my family for listening to my banal and, likely confusing, frustrations and complaints for the last 6 years. Specifically, for my mom who never doubted I would finish and my Dad who put his resignations aside to let me become a political scientist when I left for

Bloomington 10 years ago. Also, I would like to extend bonus points to all of the above for acting as though they have any idea what I am talking about most days. I know you do not have a high degree of interest or knowledge but you play along very well. Thanks Mom, Dad, Audrey,

John, and Dorothy.

ii

TABLE OF CONTENTS Acknowledgements ……………..……………………………………………………………… ii List of Tables ………….……………………………………………………………………….. v List of Figures…………...……………………………………………………………………… vi Abstract………….…………………………………………………………………………….... vii

Chapter One: Why Scandals Matter and How They Are Defined 1. INTRODUCTION………………………………………………………………………... 1 2. WHAT IS SCANDAL……………………………………………………………………. 6 Chapter Two: Scandal’s Influence on Policy and the Public 1. INTRODUCTION TO THE LITERATURE…………...... 17 2. LOBBYING REGULATIONS………………………………………………………… 18 3. THE ADOPTION OF LOBBYING LEGISLATION………………………………….. 27 4. SCANDAL: THE ENEMY OF A HEALTHY DEMOCRACY………………………. 34 Chapter Three: Data Collection and Use 1. INTRODUCTION……..………………………………………………………………. 44 2. SCANDAL AND POLICY ADOPTION DATA..……………………………………. 44 3. CANADIAN ELECTION SURVEYS………………………………………………... 58 Chapter Four: Provincial Foul Play: Political Scandal and the Canadian Provinces 1. INTRODUCTION………………………………………………………………….….. 67 2. THEORY………………………………………………………………………………. 70 3. METHODS……………………………………………………………………………... 75 4. RESULTS……………………………………………………………………………… 80 5. DISCUSSION ………………………………………………………………….……… 83 Chapter Five: The Sponsorship Scandal and Public Opinion 1. THE FEDERAL SPONSORSHIP SCANDAL ……………………...... 91 2. THEORY……………………………………………………………………………..… 92 3. METHODS……………………………………………………………………………... 97 4. RESULTS…………………………………………………………………………….. 102 5. DISCUSSION……………………………………………………………………...... 106 Chapter Six: Provincial Scandal and Public Opinion: A Look at the 2004-2008 Panel Data 1. THE FEDERAL SPONSORSHIP SCANDAL ……………………...... 112 2. THEORY…………………………………………………………………………..….. 113 3. METHODS……………………………………………………………..…………...… 116 4. RESULTS…………………………………………………………………………..…. 122 5. DISCUSSION……………………………………………………………………..…... 125 Chapter Seven: Conclusion of Findings and Future Questions 1. CONCLUSION……………………………………………………………………..… 134 2. QUESTIONS FOR FUTURE RESEARCH…………………………………………. 137

Appendix One: Cox Model Results………...…………………………………………….….. 141 iii

Appendix Two CES Questions Used.………………………………………………………... 142 Appendix Three: Codebook..……………………………………..………………………….. 144

Bibliography….……………………………………………………………………………… 145

Vita ………………………………………………………………………………………….. 160

iv

LIST OF TABLES TABLE 3.1……………………………………………………………………………. 61

TABLE 3.2……………………………………………………………………………. 62

TABLE 3.3…………………………………..……………………………………….. 63

TABLE 4.1……………………………………………………………………………. 87

TABLE 4.2……………………………………………………………………………. 88

TABLE 4.3……………………………………………………………………………. 89

TABLE 4.4……………………………………………………………………………. 90

TABLE 5.1……………………………………………………………………………. 108

TABLE 5.2……………………………………………………………………………. 109

TABLE 5.3……………………………………………………………………………. 110

TABLE 5.4……………………………………………………………………………. 111

TABLE 6.1……………………………………………………………………………. 130

TABLE 6.2……………………………………………………………………………. 131

TABLE 6.3……………………………………………………………………………. 132

TABLE 6.4…………………………………………………………………………….. 133

v

LIST OF FIGURES FIGURE 6.1………………………………………………………………………. 128

FIGURE 6.2...……………………………………………………………………. 129

vi

ABSTRACT

Political scandal is a topic that receives wide media coverage and speculation.

While political scandals have tendency to set off a firestorm of press coverage, scandals also may have great implications for democracy. A common conversation associated with political scandals has been that of lobbying reforms to reduce undue influence over the policy process. Scandals may also lead to negative feelings about political efficacy and confidence in government institutions which can threaten the democratic health of a polity. Using a combination of original data gathered from provincial news sources in

Canada and public opinion from the CES, this study reaches several conclusions. The first conclusion is that political scandals increase the adoption of lobbying policies in

Canadian provinces. Political scandals, such as the groundbreaking Canadian

Sponsorship Scandal, are also found to have deep implications on political efficacy, confidence in government, and feelings about politicians. Finally, this study suggests that while provincial-level political scandals play an important role in provincial adoptions of lobbying policy, provincial scandals do not impact evaluations of federal politics.

vii

Chapter One

Why Scandals Matter and How They Are Defined

Introduction

Few events in Canadian political history have captivated political scientists, journalists, politicians and the public as much as the Sponsorship Scandal. This unprecedented scandal involving Members of Parliament, the bureaucracy, party leadership, and even the sitting Prime Minister ended the Liberal party’s era of dominance and established a new era of politics long-awaited by Stephen Harper and his

Conservatives. The Sponsorship Scandal was a visible moment in politics which is credited not only with these striking changes in party politics but also serious changes in lobbying regulations in Canada. Since the Sponsorship Scandal was associated with unethical dealings, kickbacks, and special interests, the new Conservative government moved to reform existing lobbying statutes in quick fashion. There was a need to guard against another such scandal. Subsequently, many provinces, also aware of the media circus surrounding the Scandal, responded by adopting new and more stringent regulations for lobbyists.

The effects of the Sponsorship Scandal were wide-ranging and worthy of academic discourse. The party system was changed greatly. While the Liberal Party has now received new life under one Justin Trudeau, it still has been forced to answer to its past indiscretions and was almost unrecognizable for many years. Likewise, the Gomery

Commission, created to investigate the scandal, made staunch recommendations about

1 how to prevent another such event from happening in Canada. The federal government and provinces both used these recommendations to fashion new regulations to guard against special interests. Overall, Canada’s Sponsorship Scandal and the fallout from it is worthy of deeper inquiry. This dissertation will investigate how lobbying regulations have changed across Canadian provinces over time and will determine how political scandals, and in particular the Sponsorship Scandal, may impact public opinion.

While this dissertation advances the discussion around political scandal and lobbying regulations, the topic of political scandal is not a new one. There is a large literature on the topic. However, in this analysis, a scandal is defined as an event involving the media and political actors which debates the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. In the case of the

Sponsorship Scandal, the media event that was the Scandal highlighted a number of conflicts of interest and kickbacks between the Liberal Party and Members of Parliament

(MP’s) and the Prime Minister. Money was handed to organizations and various third parties, often which were friends of the Liberal party or pet projects of the Prime

Minister, in a way that was deemed to be unethical. This dissertation will allow for a clearer definition of scandal and will illustrate how the Sponsorship Scandal and other similar event in the Canadian province impacted policy adoption and public opinion.

Much scholarship has been written from both an empirical and normative bent on the effects of these events and their importance to public policy development. While they can be small or widespread, scandals can lead to exciting, much-debated media events which can be with various approaches from many disciplines. Scandals are ever-present in most countries and localities and, thus, provide a constant stream of cases allowing for an easy study of the media coverage surrounding such events or fodder for research on public opinion polling.

Scandals are a particularly “juicy” area of political scholarship but they are also important from both an empirical and normative standpoint. Empirically, scandals have been shown to be catalysts for change. In the case of the Sponsorship Scandal, big changes to party dominance and federal regulations of lobbying were the result. In other cases such as in the case of the U.S. Congress, scandals have led to decreases in political turnout and political efficacy. Scandals can bring about the proposal and adoption of ethics legislation, lobbying legislation, specific regulations for different sectors of the economy, and new public policies for certain agencies. This allows for the study of the policy adoption process. Secondly, as scandals are often wide scale events which receive a great degree of media frenzy, the public is largely aware of these happenings. In the case of the Sponsorship Scandal, citizens were not only able to see the fallout publicly from this event but also could read about the Gomery Commission’s findings and proceedings in their daily newspaper. Canadians also could call for change electorally by voting the Liberal Party out of office and could demand change for ethics and lobbying regulations both at the federal and provincial level.

Scandals of this magnitude are hard to ignore in the minds of voters. Because voters may see the presence of one scandal as a looming threat which requires a policy solution, a common place to put blame is on the influence and involvement of lobbyists in politics. In the Canadian case, as in the American case, the system of interest group pluralism is an essential part of the policy process. A number of groups should be at the table representing various viewpoints. In this view, interest groups play a necessary and vital role in politics and lobbyists do important work to ensure all arguments are represented. Still, there are concerns about how this works in practice, as in the

Sponsorship Scandal, it may be the case that lobbyists exercise undue influence over legislators or are involved in corrupt practices. When large, public scandals occur, voters may demand more regulation of interest groups and lobbyists. This was the case in

Canada.

Regulatory policies which seek to curb inappropriate lobbying activities and increase transparency are seen as beneficial it the minds of voters because they are supposed to prevent events like the Sponsorship Scandal from ever happening again.

However, lobbying policies are different from most types of regulatory policies. To begin with, lobbying regulations are wide-ranging and cover lobbying across policy areas. From health policy to oil drilling, lobbying regulations in places like Canada and the United States have a chance to have a much broader impact on the balance of power in a number of policy areas through regulating how lobbyists can engage in politics. This is different, for example, than environmental policy where the impact is limited in many ways to just oil drilling or safe water.

Likewise, when passing lobbying regulation, legislators are required to act against their own best interest. By passing legislation, they can appease constituents but also limit their own political power considerably. Legislators may act to ensure the terms of these regulations are favorable rather than hoping that the public and interest organizations will permit them the same benefit of the doubt. This area of policy also differs from most other areas of regulation as they regulate the legislators responsible for their passage rather than an area of the private or public sector which is not directly related to the legislature. Scandals provide a unique opportunity to study the impact of major political events on public opinion and their subsequent public reactions. This study will add to the current scholarship in political science by adding empirical results about policy adoption and public opinion in Canada and the Canadian provinces to the current discussions in comparative politics about political scandal.

This dissertation will begin by carefully defining political scandal. Using this definition, it is possible determine how political scandals across Canada have impacted the policy adoption of lobbying regulations. In the next chapter, this dissertation will then address the Sponsorship Scandal’s impact on public opinion. After all, the

Sponsorship scandal was a visible moment in Canadian history that implicated party leaders. It should be expected that this event should have an impact on public opinion across Canada. Finally, this analysis will determine if provincial-level scandals may also impact public opinion about federal politics. The results of this dissertation suggest that two linkages exist here. First, there does appear to be a linkage between political scandal and the adoption of lobbying regulations. Second, there is a relationship between scandal and negative views of government and personal political efficacy. This dissertation strongly contributes to the existing literature in comparative politics on political scandal by using original and secondary data to establish these connections. The following sections will more outline what is political scandal and how it is expected to impact public policy and public opinion in Canada.

What is Scandal?

The amount of scholarship on scandals and politics is large. As shown in the

Sponsorship Scandal, scandals can have widespread impacts on public opinion, policy passage, and party politics. Because of these important implications, scandals have become an important topic of study in a number of disciplines outside of political science such as journalism or political communication. While there is no dearth of publications on the topic of political scandal, there are also important challenges in studying scandal.

Central to these challenges is the basic issue of defining what a political scandal is. “Scandal” can come to mean an extramarital affair between a President and an intern

(Kiousis 2003) or it could be the use of one’s expense schedule to pay for private moat cleaning (Allen 2010). Both are real types of “scandals” as detailed in the political science literature but these scandals may have different implications for public policy and public opinion. For this reason, it has been difficult for studies and scholars to agree upon a standard definition of “political scandal”.

To avoid having to identify a standard definition of scandal, many studies have simply focused on one specific case study of a scandal. Examples of this involve

Congressional scandals (Dancey 2012, Peters and Welch 1980, Shea 1999, Welch and

Hibbing 1997), state-level scandals in the American case (Atkeson and Partin 1995,

Carsey and Wright 1998, Chubb 1988, Cohen 1983, Cohen and King 2004, Ulbig and

Miller 2012), American presidential scandals (Chanley, Rudolph, and Rahn 2000; Kiousis

2003; Miller 1999; Shah, Watts, Domke, and Fan 2002), and scandals in comparative context (Curtice and Jowell, 1997; Lee, 2014; Newton, 2006; Pattie and Johnson, 2001). By not creating a standardized, accepted definition of scandal, the generalizability of such studies can be questioned. After all, while “scandal” may have changed public opinion for years to come in the Watergate Scandal (Hoogenboom 1978), there is no guarantee that the Expense Scandal will have a similar impact on British public opinion. While it is known how the Lewinsky scandal “backfired” on conservative Republican members of

Congress by helping approval ratings of the Clinton administration in some ways (Shah et al. 2002), it is unclear if this is always the case.

Moreover, a definition of political scandal which is not case-specific provides improvements in external validity. Definitions which are created based on one case alone may not work appropriately when applied across states or countries or even time periods.

By using a more universal definition of scandal, this dissertation improves the external validity of conclusions. A measure based on one case study may not properly measure the concept across political scandals or localities. A good definition should be specific enough to properly identify scandals but should be general enough to be used in other cases reliably.

A definition of scandal is not just important due to questions of external validity but also because scandals are central concepts in research designs for scandal and policy adoption. As will be illustrated later in the explanation of the literature on comparative political scandals, it is important to determine which types of political scandal are most impactful on the policy process. Not all scandals can be treated equally or have been found to have the same impact the policy process. Sex scandals, as found in the American politics literature, may differ significantly from other types of scandals and actually may not directly influence policy agendas in the same way other political scandals do (Shah et al. 2002, Bhatti et al. 2013, Smyth and Taylor 2003). Leaving the U.S. case, in France sex scandals were viewed differently by voters than scandals involving bribe taking or conflicts of interest (Sarmiento-Mirwaldt et al. 2014). How one defines scandal may have large implications for the results seen.

Sex scandals are also rarely an easy target for regulation. Bedroom habits are often seen as a personal matter. They may reflect poorly on the behavior of politicians but they do not directly impact the performance of one’s duties directly and do not waste public funding in most cases. For example, a politician who chooses to have an extramarital affair may be asked to resign by his or her constituents or may be voted out of office but this does not usually signal a larger problem with democracy or government.

Meanwhile, if a legislator was found to be hiring call girls and was using public funds or election funds to pay for such expenses, this could lead to calls for tighter campaign finance regulations or legislation requiring greater transparency in public spending should be. Sex is a difficult moral quandary to regulate and few pieces of legislation have sought to do so. Only recently, a bill in the 2016 Session of the Missouri House which would require state legislators to disclose any sexual contact they had with lobbyists or members of interest groups as it would be considered a “gift” was proposed but not passed (Krieg 2016). These regulations are rare and would likely be difficult to enforce.

Including sex scandals in a study of political scandal may lead to null results while a definition of scandal which leaves sex scandals which are known to be seen differently by voters may lead to strong results. For this reason, a definition of political scandal which focuses primarily, or to a large degree, on such moral scandals is outside of the interest of this dissertation. Since lobbying regulation primarily regulates conflicts of interest and campaign and lobbyist contributions, political scandals surrounding these concepts should be the focus of this study. Thus, the definition for scandal used needs to adequately fit the needs of this research design and should be able to be applied easily.

Regarding the study of the policy adoption of lobbying regulations in the

Canadian provinces, this analysis will use news sources to determine when a scandal has occurred. How the news media define scandal also plays into this vision of scandal is well known but especially significant in studies which plan to use such a measure.

Thompson (2000) argues that it is necessary to consider what the media deem important when defining scandal so that a proper study of new sources can identify when scandals happen. Likewise, Lee (2014) argues the importance of media in defining what becomes public or salient to the average voter. These scholars stress that until the media picks up a

“scandal” it does not actually achieve scandal status. Thus, using a definition which allows for such an analysis is important based on the independent variable used.

Lee (2014) tries to solve the problem of determining what is political scandal is with a definition which includes policy actors and the media, stating, “A scandal is a communicative event in that it involves journalists, commentators, and political actors articulating, denouncing, and sometimes debating the wrongfulness of the transgression”(3). Branching off of Lee’s definition, the definition of scandal which this analysis will employ is an event involving the media and political actors which debates the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. This definition includes the executive, legislature, and bureaucracy as possible actors and covers a variety of scenarios. This definition does not, of course, include every type of “scandal”. Most notably, this definition excludes sex scandals, which are outside of the focus of all the substantive chapters of this dissertation. As stated before, these may have different implications for politics than a conflict of interest scandal or the assumption that legislators were participating actively in election fraud. This may be due to the fact that sex scandals fall outside of the bounds of political life in the eyes of some voters. Or, in the case of this study, sexual indiscretions are not easy to legislate against. Because the types of policy adoptions studied in this study and in most studies in political science concerning ethics legislation rarely concern themselves with regulations which would impact the choices of political actors in the bedroom.

This definition will allow this work to conceptualized political scandal as an important explanatory variable. This is, of course, is most necessary in the two chapters dealing with provincial scandal – Chapter Four and Chapter Six. However, it still applies to the chapter on the Sponsorship Scandal, as the Sponsorship Scandal met all the important markers of a political scandal under this definition. It is beyond the ability of this dissertation to apply this definition to cases outside of Canada at this time. However, it is the hope of this analysis that this definition could easily be applied to other cases outside of Canada and could be used by other scholars in the future. The next section will explain how the definition of scandal, as created above, will be applied to generate empirical conclusions and will briefly outline the analysis within this dissertation.

Chapter Four: Scandal and the Adoption of Lobbying Regulations in the Canadian

Provinces

Since the 1990s, nine Canadian provinces have adopted some form of lobbying regulation. Most have also made subsequent changes to these regulations. Currently, all provinces but Prince Edward Island have created some form of lobbying commission to police the behavior of legislators, bureaucrats, and members of party leadership. These bodies are sometimes given independent investigatory powers and are designed as separate agencies, as in the case of Ontario’s Office of the Integrity Commission, or are seen as an organization within the legislature which ensures legislators do not violate the provincial statutes on lobbying, such as the Registry of Lobbying in Nova Scotia.

In this chapter, the focus of this research is on the reasons lobbying policies are adopted. The key research question in this chapter which will be studied is “Do political scandals drive the adoption of lobbying regulations in the Canadian Provinces?” This chapter will use data obtained from a study of provincial, local, and nationally-focused news coverage on political scandals to determine if scandals bring about changes in lobbying in the provinces. A greater discussion of how these scandals are coded and how legislation was coded in this chapter is available in Chapter Three.

This chapter should add to the scholarship on policy adoption and should be related to existing studies in American politics about the adoption of lobbying regulations. Most importantly, however, it should add important empirical knowledge about political scandal and lobbying policy to the comparative politics literature. As will be shown in the literature review, few comparative politics studies actually provide empirical evidence outside of a case study why policies are adopted. In the past, this has left the causal linkages between the adoption of these policies and scandal to be weak.

This chapter hopes to advance the scholarship on this topic by providing an example which is outside the bounds of the United States.

Chapter Five: Public Opinion on Scandal in Canada

Scandal, as will be shown in the literature review, is linked to more negative views of personal political efficacy and reduced confidence in government and government institutions. In the chapter on public opinion and scandal in Canada, the

Sponsorship Scandal will serve as a case study. The research question for this chapter asks whether the Sponsorship Scandal harmed the democratic health of Canada by creating more negative views of government and the way citizens view their ability to participate in politics.

The Sponsorship Scandal was watershed scandal in Canadian history which rose to prominence between 2004 and 2006. The party central to this event was the Federal

Liberal Party, seen as a center-left party, which was found to be embroiled in a deep scandal. While the Scandal became public with large-scale hearings during the 2004-

2006 period, the original Sponsorship Program from which it gains its namesake, began in the 1990s. During this time, the federal Liberal Party, headed by Prime Minister Jean

Chrétien, attempted to guard against attempts at Québec separatism and the 1995 Québec

Referendum by including Canadian symbols in public events and festivals (Greene 2006).

These symbols of Canadian nationalism amounted to “sponsorship” and were attempts to curry the favor of Quebeckers towards Canada as a whole. However, even after the 1995 referendum, this program continued and with it, concerns about mismanagement and secrecy began to surface.

The program persisted long after its original purpose with more and more money being paid out for advertising in Québec, in particular. Many firms and organizations associated with the provincial Liberal Party received kickbacks even for work that was not performed (Greene 2006). In many cases, even if services were performed, money appeared to be donated for small, largely spontaneous “pet projects” such as small events in Prime Minister Chrétien’s riding (McCleery 2007). In 2003, Chrétien resigned and former Treasury Secretary Paul Martin took over as Liberal Party Leader and as Prime

Minister. Still, concerns about the “Adscam” or “Sponsorgate” lingered. After numerous questions were raised by journalists and independent audits conducted by government entities and private companies, Prime Minister Martin requested the formulation of a formal commission to investigate the charges against his party and the Canadian

Sponsorship program.

The judge appointed to head the Commission was John Gomery, a well-respected member of the bench from Québec. The task of this inquiry was to determine the degree to which the Prime Minister and his government participated in illicit dealings. Two reports were released by the Gomery Commission. The Phase I report (Commission

2006a) indicated that Chrétien and several members of government were deeply involved in the scandal and absolved Paul Martin from any responsibility. The Phase II report

(Commission 2006b) released recommendations for changes which Gomery believed would have prevented the Sponsorship Scandal and which may prevent another such scandal in the future. These changes included lobbying reforms which were later used as a basis for changes adopted by parliament in the 2008 Lobbying Act (Canada 2015).

This chapter uses public opinion data from the Canadian Election Studies to explore the relationship between the public opinion regarding the Sponsorship Scandal and views that respondents have towards government and their abilities to have a say in politics. It is presumed that voters who viewed the Sponsorship Scandal most negatively are less likely to rate themselves as highly efficacious and are much more likely to view both politicians and the government as a whole with less confidence. This chapter will add more to the comparative political literature on scandal and will shed important light on the effects of major political scandals outside of the United States.

Chapter Six: Federalism, Scandal, and Public Opinion in Canada

Canada, like its American cousin, has a federal form of government. While

Chapter Four will focus on policy adoption in the provinces, it will not delve into the relationship between scandal and public opinion in the provinces. This chapter seeks to combine the research in the aforementioned chapter on public opinion with the provincial scandal data developed within the chapter on policy passage of lobbying regulations.

This chapter investigates the interaction between the views of government and sub- national scandal. With regard to Canadian satisfaction with democracy and the

Sponsorship Scandal, Rudenmeyer and Nevitte (2016) argue that a main question left in the literature on scandal and public opinion deals with the ways in which sub-national and local scandals can impact national public opinion trends. The fact that certain jurisdictions more deeply involved in scandals will vary significantly is also supported by Bowler and Karp (2004)’s study of the U.S. and U.K. which showed those in districts who had a representative involved in a recent scandal showed different levels of confidence in government institutions and much more negative views of political efficacy. While the effect of the Sponsorship Scandal should be fairly constant across all of Canada, provinces may vary widely in the degree to which they are “scandal-ridden”.

This chapter, thus, answers the question as to whether those who come from Provinces with a higher degree of scandal reported in the news will show different levels of confidence in politicians, confidence in government, or external political efficacy than those from less scandal-ridden provinces.

Because it is difficult to find provincial political surveys, the provincial scandal data allows for voters to be assigned to groups based on where they are located. Thus, while the CES only exists on a national level, it identifies where a respondent is located.

A scandal percentage from the political scandal data developed within the chapter on policy passage will be assigned to a respondent based on province. Those who are from provinces with scandal, thus, should have a different score than those who are in a province which has not experienced scandal. Likewise, as panel data is available, it will be used to show changes in public opinion and levels of scandal over time. In provinces which experience greater levels of scandal over time, there should be a decrease in confidence in government and in political efficacy.

The analysis in this chapter should shed some light also on the relationship between the association of scandal with provincial and federal politicians. As will be shown in the following chapter, there is scholarship which shows that even local-level political scandals may damage the ways voters view politicians at the federal or national level. If this hypothesis is supported, there should be significant differences in the way voters in scandal-ridden areas view government as opposed to those from scandal-free areas. Currently, the only literature which has looked at this difference in public opinion has focused on scandals in the United States. While Bowler and Karp (2004), as discussed later, look at the differences in public opinion in voters who were in ridings in

Britain which experienced scandals personally and those who did not, there was no federal dimension to their study. Thus, this is still a new area of scholarship in studies of federal countries.

Including federalism in this discussion is important as Canada is a federal government. Federalism, however, may be even more important in the Canadian case than in the American to which it is often compared. Canadian provinces may experience greater levels of independence in terms of policy development. A good example of this exists in the ways that provinces can determine what services will and what services will not be covered under universal health coverage. This will be discussed to a greater degree in the literature review. Incorporating federalism and its impact should positively contribute to a new line of scholarship in the political scandal literature as few studies currently look into the ways that federal and sub national evaluations can be clouded by a scandal at one level versus another.

Chapter Two

Scandal’s Influence on Policy and the Public

Introduction to the Literature

Political scandals are valence issues which generate vast amounts of news coverage and possibly significant public backlash against politicians, parties, and institutions. The public is expected to be largely in agreement that political scandals are undesirable. Noted public policy scholar Theodore Lowi stated that scandals are particularly damaging to government and the actors of government because scandals are seen by the public as ‘corruption revealed’ and, moreover, a ‘breach of virtue exposed’

(Lowi 1988, vii). Scandals can be seen by voters as an alert that democratic representation may not be at its best. Worse still, studies may indicate that a scandal may lead voters to view other representatives and institutions not directly involved with a level of distrust. The reaction to scandals can have both positive and negative effects.

Scandals can generate new policy but they may also create problems for democracy by reducing political efficacy and trust in government. Likewise, though lobbying regulations are often viewed by voters positively, these new regulations may restrict pluralism and act as a barrier to entry in the political process. Thus, it is important to study political scandals and their implications.

The previous chapter shed light on the way political scandals will be defined for the purpose of this study and the basic goals of this dissertation but only a brief survey of the literature and theory was included. In this chapter, two main veins of political science scholarship will be explained to lay the foundation for the theories tested in subsequent chapters of this dissertation. While few studies have focused on the case of Canada alone, there is a great deal to be learned from the existing American politics literature on lobbying policy and the scholarship on comparative political behavior. The goal of this chapter is to present not only the background on each of the substantive, empirical chapters in this dissertation but also to tie these chapters together. While Chapter Four focuses on policy change and political scandal, Chapters Five and Six, test hypotheses regarding public opinion and scandal. These chapters differ in their focus but are closely related. Scandal may act as a catalyst for the adoption of lobbying regulations but these events do not occur in a vacuum. The goal of this dissertation is to discover what policy and democratic implications political scandals can bring. If scandals increase the likelihood of policy adoption, this can have greater implications for democracy.

Likewise, if it is true that scandals like the Sponsorship Scandal damages the views of voters regarding their own personal political efficacy and their confidence in government institutions, it is possible scandals have wider fallout than previously hypothesized.

Lobbying Regulations: Origination, Evolution, and the Consequences The study of lobbying regulations has attracted many American political scientists. While this study will focus on Canada and will primarily study why these policies are adopted, it is also important to better understand the previous literature on the subject. Canada, after all, is a pluralist system like the United States but with important differences. To begin with, money in politics is generally more restricted in the Canada.

Boatright argues that while there is no limit on spending of PAC's for advertising in the

United States, individual groups are limited to a $3000 donation per candidate per election in Canada (2009, 20). Likewise, groups are severely limited on how much they can contribute to individual candidates. In the United States, PACs may spend up to

$5000 per candidate, per election (FEC 2012). In Canada, however, contributions are currently limited to $1100 per candidate (Boatright 2009) which is significantly lower than in their neighbor to the South.

Very little research has been done on the connection between lobbying and parliamentary votes. This is likely due to the Canada’s Westminster system, where party strength controls most votes, leaving less opportunity for individual votes or voting in the best interest of one’s constituency (Franks 1987). Still, not all votes in the Canadian system require such party loyalty. Free or unwhipped votes may allow for more influence of interest groups or lobbyists. A few studies of Canadian lobbying exist in the policy literature suggest that parties and constituency factors have the most influence on free voting behavior (Hibbing & Marsh 1987, Longley 1999, Overby & Studlar 1998,

Overby & Taydas 2011, Weale & Bara 2012) but, as argued by Franks (1987), interest groups are suspected to be responsible for the abolishment of the death penalty, which was a historic free vote. Regardless, while interest groups in Canada may exert less influence than their American counterparts due to the nature of the Westminster party system, they may have more influence on unwhipped votes.

Still, campaign donations may not be the best way to illustrate the importance of interest group influence. Unlike in the U.S. case, Canadian candidates are picked by party organizations rather than by largely candidate-financed primary elections. Party rolls reduce the influence of money. Likewise, federal parties receive a reimbursement of

60% of election expenses back from Elections Canada during an election year (Boatright

2009, 22) making candidates less reliant on donors and interest groups. Thus, it may not make sense to compare the Canadian and American systems based on dollars spent on elections and money donated to politicians alone.

The lobbying regulation in Canada focuses partially on money in politics and partially on other types of influence such as meals out, contacts with legislative staff, and conflicts of interest. Federal regulations require all lobbyists and advocacy groups which will contact any bureaucrat, policymaker, or other government actor to register with the federal Office of the Commissioner on Lobbying. The Commissioner’s office keeps a log of all contacts between lobbyists and government officials in searchable database.

Many of the provinces and localities have adopted a similar or almost identical system.

The assumption made here is that campaign donations may be less important in the

Canadian case due to the role of parties and tougher election laws than seen in the

American states but the existence of such lobbying regulations shows the influence of so- called special interests is still of concern to provinces and the Canadian federal government.

The purpose of this dissertation is not to pass judgment on the effectiveness of such legislation or to make a direct comparison between campaign regulations in Canada and the United States. Instead, it focuses on the conditions which bring about these changes. Regardless of the role of money in politics, Canadians may observe a scandal and demand tighter controls for money in politics and lobbying as these voters may view the system with a lack of transparency or to be ripe with corruption. Thus, while it is important to understand the Canadian system differs from the American regulations, it is believed that lobbying regulations likely arise from the suspicion voters may have about the influence of interest group influence and worries of corruption in both cases. Much of the literature on lobbying which examines such policies rests in the

American politics scholarship on federal and state lobbying policy adoptions. A push for the regulation of campaign contributions and interest groups began in the early 1900s.

The Teapot Dome scandal of the 1920s damaged the public’s view of the Harding administration and implicated various members of Harding’s cabinet allegations of bribes and unfairly distributed oil drilling contracts in Wyoming (Bennett 1999). This led to the

Federal Corrupt Practices Act of 1925 which regulated election contributions (FEC

2015). Thus, since the 1930s in the United States, various types of organizations have also been required to register as lobbying organizations for activities conducted even outside an election period. However, as Thomas (1998) states, these regulations were weakly enforced and were more likely a form of “monitoring” than regulation. In the

1970s scandal once more spurred a change in federal election low in the United States.

The most notable of change is reflected in The Federal Election Campaign Acts of 1971 and 1974, which more tightly regulated campaign spending (FEC 2015). The 1974 amendments to this law, a response to Watergate, are responsible for establishing the

Federal Election Commission which is the independent commission created to regulate election contributions. After Watergate, the FEC adopted an approach which moved away from this idea of “monitoring” and into regulations which were more heavily enforced (Ensign 1997, Hoogenboom 1978, Thomas 1998).

Because of the fear of another such scandal occurring, many states also went forth and adopted similar and even stronger regulations to prevent abuses of power and unfair dealings made via lobbying violations. It is assumed, thus, that such legislation increases transparency, reduces abuses, and keeps both politicians and lobbying organizations honest. Various studies have attempted to analyze the “stringency” of regulations and how they impact the democracy in jurisdictions across the United States and, more recently, around the world.

Lobbying legislation, as laid out in the American case by Thomas (1998), in incredibly common but varies quite a bit across cases. Thomas states that lobbying legislation can be divided into four “elements”: those which deal with the actual lobbying of legislators, those which deal with the disclosure agreements and conflict of interest worries, those which deal with campaign finance, and those which deal with specialized

PAC’s. Thomas visualizes these “elements” as being weighted differently in legislation.

States have chosen various ways to ensure that lobbying is regulated within their borders.

Thus, some states will have stronger campaign finance requirements, others will rely more heavily on legislative and lobbying codes of conduct, and other states will more strongly police conflict of interest and disclosure through strong “revolving door” provisions which prevent legislators from re-entering political life as lobbyists immediately after they are either forced out with term limits or via the loss of an election.

Lobbying regulations have many variations, as shown above, and they come in many “shapes” and “sizes”. Likewise, these regulations have been adopted more recently outside of the U.S. As such, the study of comparative lobbying regulations has grown.

Studies regarding the process of adoption or the nature of regulation of comparative lobbying policy have been conducted in Australia (Keane 2008, Warhurst 1998), Canada

(Rush 1998), Central Europe and Eastern Europe (Harsanyi and Schmidt 2012, McGrath

2008), Denmark (Rechtman 1998), the EU (Coen 1998, 2007; Dur 2008a, 2008b;

Wesselius 2005), Hungary (Stark and Vedres 2012), Israel (Veksler 2011, Yishai 1998), Poland (Jasiecki 2006), and the UK (Allen 2010, Jasiecki 2006, Jordan 1998). As in the case of the United States, these regulations vary. For example, recent adoptions in

Eastern and Central Europe require only the registration of organized interests (Stark and

Vedres 2012) while the Canadian case requires even simple contacts to be documented between political figures and lobbyists. Meanwhile, other cases have no official policing of lobbyists but require MPs to document spending such as in the UK (Allen 2010). In

Canada, some provinces have adopted stringent regulations with strong provisions such as revolving door policies and searchable contact databases which even document the reason for contact while others adopted weaker, less-independent institutional mechanisms for regulating lobbying. This will be discussed in greater detail in the next chapter. As it stands, lobbying regulation continues to be an evolving area of regulatory policy which continues to gain attention over time and across countries.

Lobbying policies are not all created equal. As time has passed, most states have also adopted more expansive regulations (Gray and Lowery 1995). Where in many cases, lobbying policies required little more than voluntary registration and no policing, violators in many jurisdictions now face serious fines for not following a code of conduct or registering in a timely fashion. Various studies have been conducted to survey how

“stringent” these regulations are. Opheim (1991) designed an index based on the ways in which state lobbying regulation defined the relationship between those lobbying and policymakers being lobbied. Brinig, Holocombe, and Schwatzstein (1993) developed a ratings system that captured more factors that Opheim’s index but, as pointed out by

Newmark (2005, 183), neither of these previous ratings systems has been shown to be easily replicable. Thus, Newmark devised a more transparent and replicable system to rank American state ethics regulations which has been used again in one study by Ozymy

(2013).

With even Newmark’s (2005) technique, however, this creation of an index which creates equally weighted and somewhat subjective propositions which are then added up to determine what is a “stringent” lobbying regulation is problematic. Newmark’s scale, which runs from 0-18 allows 1 point for each aspect of regulation which a piece of legislation adopts, leading to higher scores for more stringent regulations and lower scores for less stringent regulations (Newmark 2005). While it may be possible to use these factors illustratively to show the various aspects of each piece of legislation adopted, it is difficult to discern if all these conditions should be ranked equally or if some factors should be weighted more importantly than others. This issue of replicability is problematic and leads to measurement error.

Put into practice, such rankings do allow for the study of the impact of regulations in a way that cannot be done with a simple binary measure of policy adoption, however.

This is important as the American politics literature has shown that lobbying regulations are not a “one-size-fits-all” proposition (Gray and Lowery 1997). Certain regulations are almost no barrier to entry for interest groups and change little substantively about lobbying in a jurisdiction. Gray and Lowery (1997) state that it has been argued that these lobbying regulations may damage the ability of certain interest groups or lobbyists which represent various viewpoints to enter. This is due to the time and economic costs of regulation. However, Gray and Lowery (1997) do not find large changes in the composition of interest group communities both before and after the enactment of such legislation and argue that most state regulations neither harm interest group participation nor do they appear to significantly change the existing interest group communities. In other words, these regulations are often formalities that may not change or improve democracy as the public may hope. In the comparative literature, as well, Chari et al.

(2007, 2010) find that regulations adopted in countries around the world vary significantly in both the sorts of protections they emphasize and the degree of transparency these laws bring about. The degree to which these regulations matter or even serve as a genuine barrier to prevent behavior is not well-observed in the comparative literature but has been hotly debated in American politics.

As Rosenthal (2000) argues, these lobbying regulations are created to bring about better accountability and foster quality democracy. Despite this hope, the literature which evaluates these lobbying policies is conflicted. In some cases, there are improvements

(Strickland 2014, Hamm et al. 1994) in transparency. Meanwhile, in others, there may be no improvement in transparency (Hunter et al. 1991; Gray and Lowery 1993, 1994,

1997). Alternatively, these regulations may have led to less interest representation and, possibly, a lower quality of democracy (Brinig et al 1993). The problem with stating that these regulations work or harm is that the tools to analyze and evaluate these policies usually rely primarily on indicators of interest group density and/or interest group diversity.

While it is important that there should remain a vibrant number of interest groups and regulations should be overly burdensome to small groups, especially citizen groups with fewer resources, there are other impacts that should also be observed. Strickland

(2014) argues in a study of U.S. interest groups in the states that regulations may not deeply impact interest group communities or inhibit interest group density. Rather, expansive regulations should be viewed positively as they were shown to positively increase transparency and regulation. Thus, these regulations should be seen as positive regardless of whether they increase or decrease interest group numbers or density.

Rosenson (2006) also suggests that such policies change the composition of state legislatures for better as they may lead to fewer candidates running who would be particularly risk-averse under ethics laws. In her study, she found that better campaign finance requirements lowered the number of candidates competing for state offices – specifically in the case of business owners – likely due to the concern of conflicts of interest becoming public (Rosenson 2006, 626). This, coupled with the findings of a recent study conducted on Members of Parliament in the UK which found that candidates for public office are more likely than the general public to be willing to commit “ethically dubious” acts (Allen and Birch 2012), suggest that such regulations may at least serve as an important gate keeping role to keep less-ethical candidates out of politics. Finally,

Ozymy (2010) states that better lobbying regulations may lead to more legislative independence from lobbyists, which suggests legislators may be more responsive to policy and constituent needs than the needs of special interests.

While the effectiveness of lobbying regulations may differ across cases and context and may still be debated, citizens are often largely in support of these regulations because they believe they increase transparency and make government more responsible and responsive. The public believes that lobbying regulations are helpful in ensuring the health of their democracies. This is supported in studies of public opinion about lobbying legislation in America (Common Cause 1991, Duquette 2002, Ozymy 2010) and comparatively (Chari 2010, Julliet 2004). At least in the minds of voters, these policies are positive. Thus, while more should be done to determine the effects of lobbying policies, those results are beyond the scope of this study. This dissertation does not seek to measures the impact of these regulations but, instead looks at the reason they came to be. Public cries for change, particularly in times of scandal, should bring about new lobbying legislation. This work does not focus on interest group density but rather other barometers for the health of a political jurisdiction. By focusing on the Canadian case during a time of great change in lobbying legislation, it is hoped more will be gained in the understanding of transparency and the stage of policy adoption.

The Adoption of Lobbying Legislation

What makes these policies different?

The study of American public policy has long appraised the reasons that states adopt lobbying reforms. One common assumption for this is that lobbying regulations are a fairly unique area of policy and, in particular, regulatory policy. When thinking about regulatory policy, it is common to think about laws which limit the amount and types of pollutants than can be released or limits on what financial institutions can do when extending credit. However, the regulation of the behavior of legislators and lobbyists also is considered an area, albeit a peculiar one, of regulatory policy. As

Rosenson (2003) notes, these policies often have economic dimensions similar to many other areas of regulatory policy and they are often adopted due to public or media outcries. Many jurisdictions, particularly in the Canadian case, deeply restrict the types of gifts legislators may take from constituents and lobbyists. At the federal level, regulations require each contact made or each cup of coffee given to be documented (Government of Canada 2015). Alternatively, these relationships may be economic in nature, as penalties in most American states require the lobbyist in violation of terms to pay a fine or be removed from the lobbying registry for a time. Losing this profitable incentive would also remove their source of income for not following the law. This is not dissimilar from an auto maker who routinely fails to meet emission standards being prohibited from selling cars or an oil company facing a fine of millions of dollars for not ensuring their drilling rigs were leak-proof. While regulatory policies may not completely eliminate of the profit margins of corporations or entities for not following regulations, fines reduce profit. Thus, regulatory agencies can reward those who follow the rules by not fining them while punishing violators for not following regulations.

Likewise, many types of regulatory policy occur because the public determines they are necessary to stop some sort of problem which has been identified. Many lobbying policies have originated from an electoral mandate following a political scandal.

In the American case, scholarly literature focuses on the adoption of lobbying regulations in U.S. states following scandals (Fields and Davis 1993, Gray and Lowery 1999). In

Arizona (Bullock 1994), Kentucky (Bullock 1994, Ensign 1997), South Carolina

(Bullock 1994, Ensign 1997), and Alabama (Ensign 1997), accounts detail that the main catalyst for the adoption of more stringent and effective lobbying regulations was due to a political scandal. It is theorized that legislators will act by passing new legislation to curb a problem as the cost of inaction is losing their jobs.

The relationship between legislator and policy also changes in the area of lobbying regulation. Rosenson (2003) argues that legislators are rational actors who should avoid adopting lobbying regulations and reforms because they are the ones who stand to lose the most by their own hands. Legislators benefit widely from their ability to work with lobbyists, after all. While politicians can receive rewards in terms of policy outcomes by working with organized interest groups (Ainsworth 1997), they can also receive benefits from everything from campaign funding to free cups of coffee from these groups as well. Also, politicians benefit from the information that lobbyists can provide

(Bennedsen and Feldman 2006). As politicians have limited time and resources to gain policy information, lobbyists and interest groups can provide good research on a variety of topics. These benefits are important to legislators not only because they allow for better policy outcomes but also can act as legislative “subsidies” which take pressure off of politicians from campaigning and, instead, allow them to focus on constituent needs and policymaking (Hall and Deardorff 2006). When these subsidies disappear, it can make the work of legislators more difficult. Those who cannot privately fund their campaigns or afford to hire outside staff during campaign season may be hurt the most by this loss of subsidy. This is why it has been found, overall, that the adoption of lobbying policies has been seen unfavorably by legislators (Beard and Horn 1976, D. Thompson

1995).

Policy Change: Scandals and Self-Interest

In times of trouble, legislators are forced to act against their own best interests to avoid losing their seats (Rosenson 2003, Ozymy 2013) or for fear of a referendum which would potentially provide much less friendly terms (Ozymy 2013). The event which has been credited with crystallizing the passage of lobbying regulations is a political scandal

(Ozymy 2010, Rosenson 2003, Witko 2007). Numerous studies of various cases of lobbying adoption have suggested that scandal is a defining reason for the adoption of lobbying reforms. Rosenson (2003) finds that the presence of political scandals is an important determinant of policy passage. Using a measure which ranks political scandals in terms of magnitude, she finds that wider-ranging or more salient scandals are more likely to bring about lobbying reforms. Witko (2007) also finds that scandal is in an important factor in the adoption of ethics reforms but scandal is not the only factor.

According to Witko’s work legislative professionalism, state ideology, and state culture can also contribute to why states adopt these reforms. Ozymy (2013) finds scandal to be a strong determinant, along with other factors, which brings about lobbying reforms. Thus, quantitative studies across the United States have suggested political scandals are an important determinant of lobbying policy change.

While scandal appears highly associated with the adoption of lobbying policies and reforms, a number of other factors have also been associated with the adoption of lobbying forms in the American states. First, most research accounts for the alternative regional diffusion hypothesis. This theory tested by Berry and Berry (1990), in which states that are adjacent are more likely to adopt the policies previously adopted by their neighbors, has been shown to also predict the adoption of lobbying legislation in

American states (Ozymy 2013, Rosenson 2005, Witko 2007). Policy diffusion in other areas but not in the case of lobbying policy has also been studied in Canada in regards to environmental regulations and health policy (Rabe 2005, Carroll and Jones 2000, Pomey et al. 2010) and the EU in regards to general regulatory policy (Jordan et al. 2003, Bakvis and Skogstad 2002). Thus, in cases where federalism is a factor or, potentially, in a cross-national study, it is important to include a measure which accounts for this regional diffusion. Similarly, states that are wealthier are more likely to adopt new regulations in a variety of policy areas. Alt and Lowery (1994) and Barilleaux and Bernick (2002) show that states who have more money to spend will be more favorable towards the adoption of new policies. This is should be expected in the case of lobbying legislation, according to Rosenson (2003) because these states will be most able to appropriate the funds to establish institutions and will be more apt to fund better enforcement efforts for years to come.

The presence of certain types of political culture is also associated with higher rates of lobbying policy adoption, as is seen in Ozymy (2013). Branching off of the theories of Elazar (1984)’s view of “moralistic culture”, states who are more moralistic should be more likely to appreciate transparency in politics. Ozymy (2013) finds that moralistically-minded states were more likely to adopt stronger lobbying policies. The partisan makeup of a legislature, too, has been associated with the passage of such reforms. In a similar fashion, Rosenson (2003), includes the other two dimensions of political culture developed by Elazar (1984) – traditionalistic and individualistic – and determines that both are, as expected, associated with a lower level of adoption of lobbying commissions in the American states. Similarly, it has been hypothesized that a state’s political ideology or the ideological culture of its citizens may impact the passage of certain policies. Witko (2007) finds that left-leaning states should be more apt to adopt regulations than more conservative states. Meanwhile, Rosenson (2003) does not find support for the argument that left-leaning states should be more likely to adopt more expansive lobbying regulations. Finally, other factors have been identified which may make certain jurisdictions more likely to adopt lobbying regulations. First and foremost, legislative professionalism has been found to increase the stringency of lobbying reform adoptions and their frequency of adoption (Witko 2007, Ozymy 2012). Opheim (1991) argues that more professionalized state legislatures are more apt to experience higher levels of independence from special interests and, thus, Ozymy (2013) predicts that these legislatures should be more open to the adoption of more stringent lobbying regulations.

Bureaucratic capacity is also identified in most studies as an important factor in policy adoption (Rosenson 2005, Witko 2007). Similarly, Rosenson (2005) argues that the literature around state wealth suggests that states with a higher gross state product should adopt more expansive lobbying regulations because these states are most able to afford the increased regulatory burden. States that are wealthier and have larger, more professional workforces should be more willing to adopt more regulations. Since lobbying policy is usually couched as a regulatory policy, this seems a safe assumption to make. However, state wealth and bureaucratic capacity have not been found to be an important determinant of the lobbying policy adoption in the literature thus far (Rosenson

2005, Witko 2007).

Outside of the United States, little research has been done to determine what brings about changes in lobbying regulations. As stated previously, most accounts of policy adoption in this area found in the comparative politics literature are single case studies full of rich description. Most studies do not include quantitative analyses and are have significant external validity limitations. To date, the most comprehensive, empirical study of lobbying policy adoptions outside the United States has been done by Chari et al. (2006, 2007, 2010). Chari et al. (2010) assembled the most rigorous and impressive cross-national study about lobbying in American states, the Canadian provinces, and other OECD nations. Similarly but not exactly like in the American politics studies,

Chari et al (2010) observed the level of corruption in all OECD countries regardless of lobbying policy passage status. They assessed how perceived corruption in a nation impacts its likelihood of both the adoption of regulations and the stringency of regulations (Chari et al. 2007, 2010). Applying an index to measure the stringency of policies cross-nationally and sub-nationally, Chari et al. determined that corruption appears to play an important role in the adoption of ethics reforms as reforms tend to occur when corruption levels are high in a nation.

Chari et al. (2007, 2010)’s work is limited in its ability to explain policy adoptions of lobbying regulations due to its sample. While Chari et al. find that cases adopt ethics reform at times of high corruption, they do so only at one time point, so no time trends can be accurately observed. The results from Chari et al. make some preliminary conclusions which are interesting but stop short of being truly a quantitative study. Thus, while Chari et al.’s (2010) work does address policy adoption and stringency of lobbying regulation, it does not rule out alternative hypotheses or look at the trends of adoption over time. While these scholars stop short of conducting a cross- national study which could link corruption to policy adoption, they do include some rich accounts for policy adoptions in a number of countries. Using various case studies, these scholars suggest that stronger regulations are likely to be adopted in nations with higher levels of corruption but their examination of these cross-national cases looks only at this one point in time. Their findings suggest that adoptions of stronger regulations have occurred in periods of high corruption but this is not tested in a quantitative study and no other factors are controlled for. Thus, comparatively, there is work to be done both in terms of explaining policy adoption and explaining why certain cases adopt stronger regulations.

While there is still progress to be made in comparative politics, this study hopes to generate conclusions from the Canadian case. A larger goal is to positively add to the comparative public policy literature in a quantitative, empirical way. By compiling data gathered from new sources, it is the hope that this dissertation will be able to show support for the link between scandal and lobbying policy changes in a way that has not be commonly done in cases outside the American states. Finally, by conducting a study across a significant span of time and across all ten provinces, it should be possible to include important control variables which account for key alternative hypotheses.

Scandal: The Enemy of a Healthy Democracy?

While the literature suggests scandals should lead to a greater likelihood of policy adoption of ethics reforms, political scandals may also bring about a negative or skeptical view of government by the general public. In subsequent chapters, this dissertation will explore this link between public opinion and scandal. The literature in both comparative and American political science concludes that scandal may have important implications for the way citizens perceive government, their elected officials, and their own abilities to contribute to the political process. Dissatisfaction with government may have negative political consequences for democracies, it is important to assess how in the case of both the Sponsorship Scandal and during various times of provincial scandal how Canadians are able to view their government and their ability to contribute to the political process.

The lessons learned from the Canada may apply more broadly, as Canada is an advanced industrial democracy which should allow these findings to be generalized to other similar cases. This section will discuss the comparative political behavior research on scandal and public opinion but will also include a discussion of the impact of federalism. As

Canada has a federal system and federalism has often been linked to accountability issues, it is of deep importance that any discussion of political behavior in Canada and take into account this important institution.

Negative Views of Government

A recent trend, especially in the field of comparative politics, is to investigate the views of citizens towards government. The concern is that negative views towards government, such as a lack of trust in government institutions or reduced voter turnout, may lead to negative consequences for the health of a democracy. In Europe and in

North America, voter turnout and traditional indicators of participation are on the decline

(Dalton 2002 and 2008, Butt and Curtice 2010, Blais 2000 and 2006, Putnam 1995,

2000). Thus, beyond Canada or the United States and throughout all of the developed world, certain traditional predictors of political participation such as joining a political party or voting regularly are on the decline.

Some scholars have questioned if political scandal may account for some of this change in voter participation (Kumlin and Essaisson 2011). In studies of American politics, political scandals may negatively impact views of candidates (Smyth and Taylor

2003) and may influence voter turnout either by reducing it (Callier 2010) or temporarily increasing it so that voters can hold incumbents accountable (Praino et al. 2013).

Similarly, while no direct causal link was found between the 2009 Expense Scandal and turnout in the 2011 electoral system referendum, Bowler and Donovan (2013) suggest that the Expense Scandal may have actually increased British turnout for the ballot measure. While the debate continues over the ways that scandal either affects evaluations of governments or reduces turnout, it is clear that scandals may change the views of voters and their actions in some way.

Various studies in American politics and a growing number of comparative politics studies, as shown before, have studied scandal’s impact on public opinion. When voters perceive behavior of political actors as particularly unsavory, it is reasonable to assume it may change their views of politicians. Bowler and Karp (2004) study scandal in both the U.S. and the U.K. after major political scandals. In the American case, the authors examined U.S. House Banking scandal where legislators were involved in writing bad checks (Bowler and Karp 2004). In the British case, a number of high profile political scandals were coded (Bowler and Karp 2004). In both cases, it was found that evaluations of politicians were negatively affected – especially in areas where members of Congress or Parliament were known to have participated in a scandal.

More disconcerting, however, is the fact that scandals may also have an impact on voters via their views of trust in government and its larger institutions. Regarding trust in government, Rudermann and Nevitte (2016) find that the attitudes of Canadian voters as the height of fallout from the Sponsorship Scandal reflected reduced levels of trust in

Canadian democracy. Kumlin and Essaisson (2011) find that government trust and satisfaction are negatively impacted by scandals, but they emphasize that this relationship is more complicated than previously thought. Kumlin and Essaisson’s study of European countries over three decades shows that as media reporting of scandals becomes more common, only major scandals – involving multiple institutions or multiple parties, in particular – are shown to have a deep impact on government trust. Leiter and Clark

(2015) also found this relationship to be more complicated than a simple link between levels of government trust and satisfaction with democracy in nine European democracies. Instead, their findings suggest that scandals and other external events which may alter trust matter most only when the opposition’s behavior appears “better”

(Leiter and Clark 2015). In other words, voters care most about the bad behavior of a governing party when they perceive a choice of a better-behaved opposition. These opinions about trust in government and evaluations with democracy are not limited to the bad choices of elected officials, however. Widespread allegations of corruption due to the poor behavior of bureaucrats in Spain were found connected to a decrease in trust in government by Villoria et al. (2007). Thus, scandals involving the bureaucracy, executive, and legislative branches have all been found to harm views of the government in citizens.

However, most of the research in public opinion and scandal has stopped short of studying specific implications of political scandal on institutions or the impact of specific scandals on voter attitudes. One main concern which Bowler and Karp (2004) investigated was the worry about a “spillover” effect in times of scandal where voters not only begin to have less trust in government system (such as the concept of democracy) of a country or in specific politicians, but also in specific institutions like parliament. These results show that scandals make voters more likely to lose trust in specific institutions. These results are important because while voters may feel a decrease in general notions of political trust, they also may experience deep decreases in trust in specific institutions.

An additional reaction to scandal which has only recently been discussed in the literature is that of the effect of scandal on political efficacy. Villoria et al. (2007)’s results in a study of bureaucratic scandal in Spain supports the hypothesis that scandal reduces internal feelings of efficacy. In addition, feelings of interpersonal trust were lowered during the scandal. Bowler and Karp (2004) find that those most affected by political scandal, or who have a representative from their district who was a participant in a recent scandal, feel lower levels of efficacy as well. Once more, these results suggest that scandals may have much deeper impacts than previously expected on notions of individual political power.

While these studies grant support to the idea that scandal may decrease the democratic health of a country or state, they are not without issue. As shown previously, an issue regarding the study of scandal seems to be these studies are often descriptive, qualitative, and focus on one or two specific cases of scandal at a given point in time.

This approach does not allow for the examination of alternative hypotheses. Studies with larger samples which are cross-sectional and empirical have been conducted using public opinion survey data but they also have similar issues of a lack of time trend or individual- level data. Of course, many of these problems arise due to a lack of available data – especially outside of the United States. While data from the Eurobarometer studies and the Canadian National Election study may be useful for political scientists, these surveys do not always ask the same questions every year in a reliable way. Due to this inconsistency, there are problems with the generalizability of the results. Longer time trends, as are seen in studies like Kumlin and Easiasson (2011) provide a way around this problem. If similar questions are asked in a study every 4 or 5 years for 50 years, the number of time points will allow for a significant amount of variation in events, such as scandal.

Still cross-national studies of public opinion can still have issues getting around the ecological fallacy by aggregating responses based on the mean response to a question.

This approach limits the ability to test for alternative hypotheses using individual-level factors such as political sophistication or education which could play an important role in voter conceptions of efficacy, trust, or approval of government. Descriptive or aggregate studies may be good at introducing a conversation about scandal and government trust but they do not allow for important causal assumptions to be made.

One solution to this problem involves finding cases which provide a natural experiment. Bowler and Karp (2004) break respondents into natural treatment and control groups for scandal by labeling respondents as either scandal-averse or scandal- free. This condition was based whether or not the MP or Member of Congress from their district or riding was implicated in a scandal but this requires either a large time trend in a cross-national studies or the ability to break the population into smaller geographic units via electoral or sub-national boundaries. This is perhaps possible in federal systems like

Germany or Canada or in single member district (SMD) systems like Britain but most electoral systems in industrialized nations do not fit neatly into either or both of these categories.

Finally, as will be addressed in the following chapter, another issue with studying scandals is how to identify scandals. This dissertation provides an improved method for identifying and coding scandals. Previously, approaches such as the use of source materials like the U.S. Book of the States for an American study or Keesing’s Record of

World Events, as used by Leiter and Clark (2015), have been used in cross-national studies of industrialized nations but would not be useful in a study of local elections or a study of new democracies. As Kumlin and Asiasson (2011) argue, voters can become conditioned to scandal or “fatigued” unless a scandal is truly widespread. So, it is important to characterize and distinguish between small and large scandals. A simply binary measure, as has been used in many past studies, may not accurately measure the presence of scandal.

The literature on public opinion and scandal is evolving and has progressed over time. Simple accounts of candidate evaluations were only the beginning of what is now a complex study of voter behavior and opinions throughout the world. While the literature diversifies, it still faces issues concerning data availability and a limited number of techniques which can be used to analyze the effects of public opinion and scandal. While this study will not be able to overcome all of these challenges, it does attempt to address many of them. Especially by incorporating the federal experience of Canada in the evaluation of scandal’s effect on public opinion, it is the hope that this dissertation will advance the study of political scandal in comparative politics.

Scandal and Federalism

As stated previously, a fundamental issue with the literature on political scandal is with assessing political scandals that happen at the local or subnational level. Canada is a federal system where scandals occur at all levels of government and may interact. As such, more needs to be done to explore how subnational scandals may impact not only views of provincial politics but also federal politics. Recently, in the discussion of their results from a study of the Sponsorship Scandal, Ruderman and Nevitte (2016) argue that more investigation should be conducted to determine how political scandals at the provincial and local levels may also impact the way that voters perceive the federal government. While no such empirical examination of this phenomenon currently exists, there is some literature which may shed some light on the relationship between federalism and the impact of political scandal.

Ulbig and Miller (2012) found a link between scandals at one level of government impacting the views of actors at another level of government. Ulbig and Miller conducted an analysis of U.S. states which either did or did not see political scandal in the governor’s office. They used public opinion data to determine as to whether those people living in states which experienced scandal also judged the President more harshly.

This means that scandals at a local level can lead voters to view other government leaders at different levels of government which did not experience scandal more negatively.

Controlling for media and political knowledge, they found that while gubernatorial approval ratings suffer most from state-level scandals, the presidential approval ratings of people in a state experiencing scandal are also lowered. Scandal also appears to have an impact on these evaluations.

Studies similar to those above have also been conducted cross-nationally and in the case of Canada. While more focused on “federal accountability” or the ability to

“throw the rascals out,” these studies illustrate the issue that federalism poses to voters and citizens in a polity. Downs (1999) and Lewis-Beck (1988) find that in European nations, voters are less-apt to “blame” the proper level of government or hold the proper officials accountable for actions in federal systems than in unitary systems. In a study of

Argentina, Gélineau and Remmer (2005) found that all levels of government are blamed for national economic problems in elections, further supporting the argument that federal systems may see lower-levels of accountability and proper blame-attribution.

Most specifically, in Canada, voters often act similarly to voters in the United

States or in other industrialized federal nations. In studies of various salient policies areas such as health care, Cutler (2004, 2008) found that voters struggled to “blame” the appropriate level of the Canadian government for social programs. These voters should have been easily able to attribute “blame” to the appropriate level of government which administered the program. However, voters were more likely to appropriate blame in matters of the economy (Cutler, 2004). Confounding results are found by Gélineau and

Bélanger (2005). They found Canadian citizens had trouble attributing blame and accountability to the proper level of government when it came to economic problems.

Based on this short summary of studies, it can be suggested that Canadians are similar to

Americans in terms of being able to “blame” and have difficulty attributing “who did what”.

Thus, as seen by Ulbig and Miller (2012), it is reasonable to imagine a situation where blame for who is actually responsible in a scandal is difficult for voters to comprehend. Scandals are an important issue to study because when they occur, they are often highly-salient. Powell and Whitten (1993) and Cutler (2004, 2008) have, in the past, stressed the importance of studying salient issues in studies of federal accountability. Thus, while scandals may not be as constant an issue as economic accountability, they may often be in the forefront of many voters’ minds at the polls. For this reason, the next chapter will explain how political scandals were identified in

Canadian provinces and develop a system of analysis which will assess the impacts of scandal on public opinion and public policy.

Chapter Three

Data Collection and Use

Introduction

Data for this dissertation were compiled using a variety of sources. Chapter Four uses a primary data source which was gathered while using archival newspaper sources and provincial statutes. Chapters Five and Six use secondary data gathered during the

2004-2006 wave of the Canadian National Election Study. Each dataset comes with its own benefits and drawbacks. This short chapter will explain how original data were gathered, coded, used, and what the challenges are with employing data in this manner.

While the empirical chapters of this analysis are linked in subject matter, the different research questions require different approaches in analysis. And while some data will be used in multiple chapters, it is important to discuss these sources separately.

Scandal and Policy Adoption Data

The research question behind the empirical test below is simple: do provincial scandals make the adoption of lobbying reforms more likely? While it may be easy to write a descriptive account which assumes a policy adoption of lobbying registrations was related to a recent political scandal, empirically assessing this is difficult. There are few empirical studies on this topic – particularly outside the United States. Most studies rely on a single case study. This method of research may make it difficult to assess causality. Chapter Four assesses this relationship across the Canadian provinces and over a substantial time period. This time period is necessary for an empirical, quantitative approach to be successful. Unlike the United States, Canada only has 10 provinces. By using a time series method, the sample size should be sufficient for a statistical analysis.

Data for manipulation were gathered originally. The dependent variable, policy adoption of lobbying legislation, and the independent variable, political scandal, were created via the coding of original sources. Prior to this original data creation, there were no preexisting measures of scandal in Canada or the Canadian provinces and no measures of changes in lobbying legislation. Instead, these data had to be gathered through a content analysis and close reading of lobbying statutes in the Canadian provinces.

In order to conduct a study of the political scandal and lobbying policy, it was imperative to be able to first find scandals at the provincial level. No such measure existed. There was no listing or central database of provincial scandals in Canada which could be used as a proxy for this independent variable. Likewise, data on the adoption of lobbying policy in Canada by year were not available. This analysis required a large amount of original data generation for this dependent variable by reading individual lobbying statutes for changes by province by year.

Independent Variable

Data collection was made possible, in part, with a grant from the Kinder Forum on Constitutional Democracy. The grant subsidized field research in Ottawa, Ontario.

Another grant of support came from the University of Ottawa. The University of Ottawa provided office space, access to archives, library sources unavailable at the University of

Missouri, and regular meetings with fellows and scholars familiar with political scandals and public policy. After meeting with Canadian public policy experts, recommendations were given regarding which news sources and databases may be useful for this study.

Using these resources, a list of news sources was generated and news articles were collected over a 20-year time period. Without the support of the Kinder Forum on

Constitutional Democracy and the University of Ottawa, it would not have been possible to gather data in the way which was required.

News sources will be used to generate a variable which can proxy for the level of political scandal. As shown in the literature, there are various ways to code for scandals.

Many approaches such as the use of source materials like the Book of the States or

Keesing’s World Events to generate lists of scandals for American states or advanced industrial states. However, in the Canadian provinces, there is no such record of scandals available. For example, Keesing’s would likely code national scandals such as the

Sponsorship Scandal but would not account for provincial-level scandals such as the allegation of vote-rigging in British Columbia lodged against an a Member of a

Legislative Assembly (MLA). News sources allow for a gauge of the public’s knowledge of scandals because they reflect what the public sees. While the news media may not cover every scandal widely or even at all, news coverage is a good proxy for citizen knowledge of scandal.

Citizen knowledge of scandal should explain why provinces should adopt lobbying legislation. It is due to the awareness of scandal that citizens should be compelled to act. In the case of political scandals, citizens are expected to act by putting pressure on policymakers to enact reform. These MLAs, fearing retribution, are hypothesized to act against self-interest and enact laws limiting their own powers to stay in office. Using news sources to gauge scandal is not a perfect science but it does illustrate when the public should be most aware of scandal and should be likely to pressure MLA’s into changing lobbying regulations in a province or adopting a province’s first lobbying reform. Newspapers will be used for this variable rather than digital or video media because they were easily searchable and provided rich accounts of scandals. Having a detailed account rather than a small sound bite or a short news report from a content aggregator was advantageous because the definition of scandal used required more detailed knowledge of a particular event. While the word scandal could be used in a news story on television or a short blurb published by Huffington Post Canada, the detail required to determine whether this event met the definition of scandal may not have been available.

Using news sources to proxy for scandal is not unprecedented in the literature.

For example, Ozymy (2012) created a measure of scandal which was coded from 0 to 1 where “0” equaled no scandal in a given state and “1” equaled a scandal occurring in a state during a given year based on a content analysis of newspapers over a ten year period. Ozymy then used this in his analysis of why states adopt lobbying reform. Thus, in the literature, similar measures have been used. Newspapers may not be a perfect source for political scandals but they do reflect the information in the minds of citizens which should frame the debate and advocacy efforts for lobbying regulations. Ozymy

(2012) also states that another benefit to using newspapers is that it is possible to create a sample covering a large number of years of policy adoption. The sample used in this analysis covers a period from 1995-2014. This 20 year period can easily be found in newspaper sources which are easily searchable via online databases. Ease of accessibility for a large period of time is another reason this is a method academics like Ozymy prefer to use news sources.

Still, even though it is clear there is a benefit to using newspapers to proxy scandal, choosing the proper new sources requires knowledge of the type of political scandals which should be included in the sample. When choosing sources, it was important to find sources which would cover political scandal at the local level. Detailed accounts were necessary to properly apply the definition of political scandal. Local sources which were more focused on issues impacting citizens at the local and provincial level were preferred for this reason. National newspapers such as the Globe and Mail and National Post tend to focus on federal politics or the provincial politics of large provinces such as Ontario or Quebec. These newspapers would not sufficiently cover a scandal in Charlottetown, PEI or Winnipeg, Manitoba. Thus, local news sources were preferred to convey an accurate, detailed account of provincial scandals. The sources used are reported in Table 3.1 below:

[Table 3.1 about here]

Included are a number of local sources in various provinces but also some national news sources or newswires. Whenever possible, local sources were primarily used. In years where coverage was not available for provinces, national sources such as The Canadian

Press, The Globe and Mail, and CanWest News Service were allowed to supplement the search.

The time period of 1995 to 2014 encompasses time covered by digital newspaper archives accessible at the University of Ottawa, which had a large number of Canadian newspaper sources not accessible via the University of Missouri’s library system. These years allowed for coverage of all provinces from 1995 onward. Because not all news sources had 2015 editions available at the time this analysis began, the dataset is censored at December 31, 2014. Still, this procedure provided 200 separate observations and ensured that all provinces had coverage for every time point in the dataset.

The news database search criteria were adapted using a number of preliminary searches in each province over time. The primary database used was Factiva via the

University of Ottawa’s library. At the University of Ottawa, a particular collection used was the Canadian Newsstand, a digital resource which provides primarily local news coverage and some larger news coverage through the Factiva database. While this collection was good at providing local newspaper articles from various local sources, as is seen in Table 3.1, this collection did not provide a local source for every province over the entire sample. In order to supplement in certain time periods for smaller provinces such as PEI and New Brunswick where local sources were only available for a short period of time, national newswires and newspapers available on Factiva was used as well. These sources did include the The Canadian Press, Canwest News, The Globe and

Mail, and The Toronto Star which had a more nationally-focused market. Including these sources, provided more coverage for smaller provinces. While it would have been preferred to use all local sources for all years in the sample, this was not possible, and relying only on provincial-focused sources would have led to large holes in data.

Over time, a set of search terms for political scandal was established as follows and was used across these two databases:

(corruption OR scandal) AND “[name of province]” AND provinc* This was found to be superior to any other search criteria used because it contained a manageable number of articles for each province without being too narrow. Using specific terms such as “legislator” or “MLA” or “assembly” were also included during preliminary searches but these searches included an incredibly large number of unrelated articles. Using “legislator” in the criteria, for example, led to a sample of over 20,000 articles in large provinces such as Ontario or Québec. Very few of these articles were related to “scandals” let alone “political scandals”. The term scandal is used often to describe a number of occurrences which are not even tangentially related to political scandal. For example, a legislator may have had an extramarital affair that became public. This did not directly relate to the definition of “political scandal” as applied in this dissertation. Likewise, various events such as citizens not getting good constituent service may be titled “scandals” but may not meet the definition. Thus, this sample would have inflated the number of articles which could have been coded by one person over a short period of time. Adding corruption to the search terms proved helpful because, while corruption has its own distinct definition concept in political science, journalists often use it to describe any type of political scandal. The inclusion of

“provinc*” in this criteria was essential as, without the limitation, a significant number of extraneous federal results were found as opposed to important provincial results.

In the end, more than 3,000 articles were coded. Table 3.2 shows the number of articles coded by province and also the percent of articles coded as scandal by province:

[Table 3.2 about Here]

Clearly, there is a disparity between the number of articles in some larger provinces and small provinces. This problem seems to be exacerbated by a lack of press coverage which is digitally available for smaller newspapers. While University of Ottawa had substantial news resources it had few sources for the Maritime provinces, in particular.

After consulting with a number of graduate students from Newfoundland, Nova Scotia, and New Brunswick and librarians at Ottawa, it was agreed that these small newspapers were difficult to access digitally and that short of travelling there physically, it would be impossible to enlarge the sample using local sources. This is why national newspapers and news databases were used. While including national newspapers in the sample did provide more articles in smaller provinces, news coverage in larger, more populous provinces like Ontario and Québec was still greater.

Only English-speaking news sources were included in this sample. Some provinces, such as Québec and New Brunswick have a substantial French language population which may get its news primarily from French language newspapers. It is possible types of articles coded in the case of Quebec may differ significantly from the

Francophone articles published. This may be due to different types of political coverage between French and English news sources (Antecol and Endersby 1999). While an attempt was made to find articles in French language sources, there was a limitation in this researcher’s ability to understand Quebecois French sufficiently. As time was an important restriction and accuracy was prized more than French news sources, the

Anglophone source The Montreal Gazette was the choice for Quebec provincial news.

Thus, the coverage may differ slightly in the types of scandals covered between the

English or French news or may be framed differently, but scandals should be covered in both types of newspapers. In both NB and QC, there were articles coded for scandal. If more time or resources were available, it may have been possible to include Francophone coders but this was not an option during this timeframe.

Only articles describing provincial political scandals which meet the definition of political scandal which is defined as an event involving the media and political actors which debates the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. These articles were coded as a “1” for scandal. Also included in this coding scheme were the political actors involved: the Premier, provincial assembly members, the government, party leadership, and the civil service.

A contrasting example of this coding scheme comes from New Brunswick. In

2004, a number of “scandals” were written about by the Telgraph Journal’s political reporter Kathy Kaulfield but most did not meet the definition of political scandal used here. For example, in an article from 2004, Kaulfield writes about a riding held by Andy

Scott, a Liberal member of provincial cabinet, who was recently caught up in two

“scandals” over his support of gun control and same sex marriage. This is a colloquial use of the word scandal and would not rise to the definition of scandal used within this analysis. Meanwhile, later in October, Kaulfield (2004) addressed the SEED Scandal which was quid-pro-quo attempt by a Progressive Conservative MLA John Betts. It was suspected that Betts required organizations applying for job training co-op grants to attend his charity event (Kaulfield, October 19 2014). While it was later found that Mr.

Betts had encouraged such an agreement, the Conflict of Interest Commissioner said he did not violate the current statute and recommended a change to existing statute which would include such conflicts of interest in the future (Kaulfield, October 19 2004). This scandal did meet the definition of a political scandal as it involved a conflict of interest which could have furthered Mr. Betts’s charity’s financial goals.

Another type of scandal which made up a bulk of the articles found but ultimately not coded as political scandal came from political scandals outside of the province. A main reason for a number of reports on this issue was due to the reaction locally to the federal Sponsor Scandal. Coming back to New Brunswick, for example, the provincial elections were held soon after the break of the Sponsorship Scandal. While these issues were not central to concerns in New Brunswick politics, many editorials and news articles were written with the tenor of a worry for democracy. For example, while some news accounts simply covered which parties were involved (Black 2004) others were written to urge voters not to trust the provincial government of New Brunswick with their tax dollars. An anonymous editorial from the Telegraph-Journal touts,

“All one has to do is look at the billion-dollar cost overrun on the federal gun registry or the $100-million federal sponsorship program scandal to know that it is never a good idea to get the government involved when it comes to having control of the tax dollars you and I pay” (Anonymous April 8 2014).

Other articles, still talked about political happenings not only in federal politics but also in other provinces. Black (2004) covers the Sponsorship’s impact on politics in Canada, for example. While the Sponsorship Scandal meets the definition of political scandal per this analysis, articles were not coded referencing this scandal alone. As only articles were coded regarding scandals which were provincial in nature and for the province in which they were written, these articles were not coded as being related to political scandal. The type of narrative included was also coded – episodic or thematic- because these are concepts which may be useful in future studies if this data is to be used by other scholars in the future. The issue of episodic vs. thematic framing was highlighted by

Iyengar’s (1991) analysis of television news stories. Episodic stories describe events or explain phenomena through the lens of a one-time descriptive account rather than in the context of history or as one event out of many. Thematic stories describe events within historical context and usually explain events by examining the past facts. These two types of framing were coding. About 60% of the articles coded showed episodic framing and about 40% exhibited the characteristics of thematic framing. As Iyengar argues that episodic accounts of important topics may reduce them to a series of anecdotes without context and may damage journalistic credibility (1991). More investigation is needed into this particular topic but the conclusion that episodic framing dominates this sample may suggest that the coverage of scandal is more likely to reduce political scandals to sound bites which may not provide the same information that a thematic account would provide.

This content analysis allowed for the generation of two conceptions of the independent scandal variable. One iteration was a scandal binary variable which was just a simple measure of whether or not a province experienced any scandal at a given time.

If any articles were coded for scandals for a province in a given year, this would be coded as a “1”. Otherwise, provinces were coded as “0” in a given year for having no incidences of scandal. The second variable creates an index which divides the number of articles coded for scandal in a province in a year divided by the total number of articles found to mention “scandal”. This can be seen in Table 3.3. [Table 3.3 about here]

Every province had at least one scandal but some had higher numbers of articles coded than others. For example, Newfoundland and Labrador experienced many incidences of scandal in several years, driving their numbers up. In this particular case of political scandal in Newfoundland, the scandal permeated all provincial institutions and was visible – consuming most of the media coverage in that province for several years. The total number of articles coded by province and coded as political scandal by province are available in Table 3.2. There is definite variation over time in scandals and this allows for a discussion of not only the impact of scandal but also the degree to which having more scandals leads to a higher likelihood of policy adoption. This variable also allows for a test of the impact of provincial scandal on views of public opinion of government and politics in Canada following the Sponsorship Scandal.

Dependent Variable

In order to understand this research design and analysis of provincial lobbying regulation adoptions in full, it is also important to illustrate how the dependent variable is coded. The indicator which accounts for the passage of lobbying legislation is a dummy variable which is coded as “1” for the passage of a policy and “0” as the lack of a lobbying policy passage in a given year. Most initial adoptions of lobbying legislation and subsequent changes to this legislation actually occurred during the study period of

1995-2014 (Commissioner of Lobbying 2014), so there are few cases of policy passage which fall out of this sample. Regardless of the “stringency” or “expansiveness” of a given lobbying statute, all lobbying laws adopted are coded as a “1”. As shown in the literature review, American politics and public policy scholars have often attempted to find a way to analyze the stringency or expansiveness of lobbying or ethics legislation but there is no yet accepted response to this problem. While some researchers claim their methods are replicable (Newmark 2005, Rosenson 2003, Ozymy 2012), this claim has not been substantiated. Due to the important concern of measurement error, all passages of lobbying regulations will be coded the same regardless of perceived differences in

“stringency” or “expansiveness.”

The indicator for policy passage is coded based on secondary sources available.

The first source used to determine when a province first adopted a piece of lobbying legislation is taken from regular reports on lobbying regulations generated by the Office of the Commissioner of Lobbying. While in Ottawa, I met with the Deputy

Commissioner of Lobbying, René Leblanc, and his researchers who agreed to send recently yearly reports (from 2012, 2013, and 2014) documenting the history of lobbying laws across Canada. These reports, which are released to the public, researchers, and shared with foreign governments and state governments when requested, detail when lobbying regulations were adopted and give general background on recent updates to the legislation (2014). These annual reports detail basic information about when, historically, changes to existing lobbying regulations have been adopted but do not provide deep accounts of changes to the law.

Another source of information which was crucial for coding this dependent variable was the use of the CanLii database to determine when exactly changes were made to these bills and to what extent. While the Commission’s reports were helpful, they did not have enough detail included to pinpoint every policy change made to these statutes over the time period of my sample. The CanLii database is publically available and illustrates any changes to legislation since passage. When lobbying regulations were passed by a provincial assembly for the first time or subsequent amendments were made to existing legislation, this was coded as a “1” in a given year. If there was no change, a given year was coded as a “0” for that province. To be coded as a change, more than just language of a law was required to change. A simple cleaning up of language which did not appear to create any new implications for lobbying would not be coded as a “1”.

However, the edition of a new online lobbying registry, such as that which was adopted in 2007 by Alberta, would be coded. Likewise, so would the adoption of an official lobbyists code of conduct, as was adopted by Quebec in 2004 .While it is not possible to ensure that all changes coded would have the same implications on lobbyists or interest groups, determining that is beyond the scope of this research.

Thus, the variable which shows the policy change includes both initial adoptions and subsequent changes to lobbying legislation. Again, there are limits to this measure.

By including both concepts in one, some distinction is lost. It also makes estimation significantly more complicated when using this measure in a survival analysis as it requires a repeated events model. Including both types of policy passage in one model, however, is important as these statutes evolve over time. It is also expected that either type of adoption should be more likely in times of scandal. Together, the original data on scandal and policy adoption are combined with a number of control variables which will be described in detail in subsequent chapters. This dataset will allow for hypothesis testing which can show if political scandal in a province increases the likelihood of subsequent lobbying legislation in given province at a given time. Canadian Election Surveys

The Canadian Election Study (CES) is a survey that is collected by established research teams across Canada during an election years. Among the 2004, 2006, and 2008 federal parliamentary elections in Canada, the CES fielded surveys to measure public opinion on a variety of topics central to these three from the federal elections. In

Chapters Five and Six, surveys of public opinion will be used to assess the impact of the federal Sponsorship Scandal and provincial scandals have on external political efficacy and the government at large.

Over the course of the 2004 and 2006 election cycles there were two types of surveys conducted by the CES. The first type of survey is the largest – the cross- sectional survey. This is a one-time survey conducted across all provinces at one time point – 2004, 2006, or 2008. This is the CES survey which has been conducted in most recent election years. This battery of studies conducted by the CES also allowed for a panel of voters and measured the public opinion of a large group of Canadian voters from

2004 to 2008. In this longitudinal study, it is possible to gather responses to similar questions across two or three elections in this period. The interviews for the above study were conducted face-to-face. Finally, the CES sent out a mail back survey (MBS), which was sent to a mailed to a smaller number of households in 2004, 2006, and 2008. Both the MBS and the cross-sectional studies included pre- and post-election waves which may have asked different questions at different points in time. CES studies are often used in similar ways in Canadian politics research to the ways the ANES studies are used in

American politics and serve as a primary source of data for existing analyses of public attitudes and opinions in Canadian politics research. Together, this dissertation will utilize these two CES studies to determine the effects of political scandal on public opinion after the Sponsorship Scandal.

In the chapter on public opinion and the Sponsorship Scandal, questions from the

MBS and panel studies will be used to investigate the relationship between political scandal and political efficacy and views of government over time. In the chapter on public opinion and provincial scandal, results from the cross-sectional study will be used to examine the effect of provincial scandal on political efficacy and views of government.

These analyses will pair the original scandal data discussed previously with the public opinion data from the CES to account for the effect of provincial scandal on public opinion following the Sponsorship Scandal. As the CES is not a survey of provincial attitudes, it alone cannot show the impact of provincial political scandals on public opinion. However, paired with the provincial scandal data, it is possible to determine which respondents should be most affected by provincial scandal and if this changes their views towards federal politics and federal political institutions similar to what was attempted by Bowler and Karp (2004).

A main issue that comes with the using the CES to pinpoint the effects of provincial scandal on federal public opinion is that large, populous provinces may dominate even this large sample of public opinion in Canada. Similar to the issue imposed by using news coverage to account for provincial scandals, small samples in small provinces could cause some degree of bias. It is possible, thus, that the preferences of populous provinces will be expressed more heavily than small provinces. However, this is the case across the board when using most large-scale surveys and the inclusion of weights should limit the impact of these biases. In the end, by including this primary and secondary data, it will be possible to determine if and how scandals influence public policy and public opinion in Canada.

Further explanations of data and methods will be included in subsequent empirical chapters. This chapter set out to better explain the process of original data collection and generation in this study. In addition, this chapter also explained and described the source of survey resource also utilized.

Table 3.1 News Sources by Province

Name Coverage Calgary Herald Alberta/Calgary The Province British Columbia/Vancouver Vancouver Sun British Columbia/Vancouver The Montreal Gazette Quebec/Montreal (English Only) Winnipeg Free Press Manitoba/Winnipeg Telegraph-Journal New Brunswick/Saint John The Guardian PEI/Charlottetown Telegram Newfoundland and Labrador/St. John's Toronto Star Ontario/Toronto The Star Phoenix Saskatchewan/Saskatoon CanWest New Service News Service/National Ottawa Citizen Ontario/Ottawa The Canadian Press News Service/National Globe and Mail National/Ontario Table 3.2: Articles Coded by Province Province Total Total Articles Percent Coded as Articles Coded As Scandal Found Scandal Alberta 285 11 4% British Columbia 252 57 23% Manitoba 251 16 6% New Brunswick 288 9 3 % Newfoundland and Labrador 196 170 87% Nova Scotia 157 74 47% Ontario 775 54 7% Prince Edward Island 28 4 14% Quebec 879 135 15% Saskatchewan 201 38 19% TOTAL 3312 568 17%

Table 3.3: Scandal Index by Province and Year Year Prov. Index Value 2005 BC 0.052631579

1999 AB 0 2006 BC 0.125

2000 AB 0 2007 BC 0.2

2001 AB 0 2008 BC 0.333333333

2002 AB 0.125 2009 BC 0.25

2003 AB 0 2010 BC 1

2004 AB 0 2011 BC 0

2005 AB 0 2012 BC 0

2006 AB 0.0625 2013 BC 0.4

2007 AB 0.055555556 2014 BC 0

2008 AB 0 1999 QC 0.047619048

2009 AB 0 2000 QC 0.04

2010 AB 0 2001 QC 0

2011 AB 0 2002 QC 0.078431373

2012 AB 0 2003 QC 0

2013 AB 0.125 2004 QC 0.161764706

2014 AB 0.2 2005 QC 0.047058824

1999 BC 0.444444444 2006 QC 0.153846154

2000 BC 0.272727273 2007 QC 0.217391304

2001 BC 0 2008 QC 0.25

2002 BC 0 2009 QC 0.064516129

2003 BC 0 2010 QC 0.333333333

2004 BC 0.1 2011 QC 0.53968254 2012 QC 0.131147541 2004 NB 0.04

2013 QC 0.183098592 2005 NB 0

2014 QC 0.380952381 2006 NB 0

1999 MB 0 2007 NB 0.090909091

2000 MB 0 2008 NB 0

2001 MB 0 2009 NB 0

2002 MB 0 2010 NB 0

2003 MB 0 2011 NB 0

2004 MB 0.1875 2012 NB 0.083333333

2005 MB 0.178571429 2013 NB 0

2006 MB 0.15 2014 NB 0

2007 MB 0 1999 NF 0

2008 MB 0 2000 NF 0

2009 MB 0.095238095 2001 NF 0

2010 MB 0 2002 NF 0

2011 MB 0 2003 NF 0

2012 MB 0 2004 NF 0

2013 MB 0 2005 NF 0

2014 MB 0.111111111 2006 NF 1

1999 NB 0 2007 NF 1

2000 NB 0.1 2008 NF 0.8

2001 NB 0 2009 NF 0.857142857

2002 NB 0 2010 NF 0.833333333

2003 NB 0 2011 NF 0.2 2012 NF 0.5 2004 ON 0

2013 NF 0 2005 ON 0

2014 NF 0 2006 ON 0

1999 NS 0 2007 ON 0.081081081

2000 NS 1 2008 ON 0.066666667

2001 NS 0 2009 ON 0.447368421

2002 NS 0 2010 ON 0.08

2003 NS 0 2011 ON 0.045454545

2004 NS 0 2012 ON 0.166666667

2005 NS 0 2013 ON 0.111111111

2006 NS 0.333333333 2014 ON 0.461538462

2007 NS 0 1999 PE 0

2008 NS 0 2000 PE 0.25

2009 NS 0 2001 PE 0

2010 NS 0 2002 PE 0

2011 NS 1 2003 PE 0

2012 NS 1 2004 PE 0

2013 NS 0.777777778 2005 PE 0

2014 NS 1 2006 PE 0

1999 ON 0 2007 PE 0

2000 ON 0.0625 2008 PE 0

2001 ON 0 2009 PE 0

2002 ON 0.029411765 2010 PE 1

2003 ON 0.055555556 2011 PE 1 2012 PE 0 2006 SK 0

2013 PE 0 2007 SK 0.333333333

2014 PE 0 2008 SK 0.5

1999 SK 0 2009 SK 1

2000 SK 0.1 2010 SK 0

2001 SK 0 2011 SK 0

2002 SK 0 2012 SK 0

2003 SK 0 2013 SK 0

2004 SK 0.1 2014 SK 0.25

2005 SK 0.047619048

Chapter Four

Provincial Foul Play: Political Scandal and the Canadian Provinces

Introduction

Lobbying is an essential political function of most advanced industrial democracies. Lobbyists and interest group organizations perform important jobs by representing citizens, corporate interests, and even government organizations. In pluralist systems like the United States and Canada, the goal is competition. Schattschneider argues that, “Above everything, the people are powerless if the political enterprise is competitive” (1960, 140). Competing interests will best represent the interests of a democratic populace. However, over time, it may seem that lobbying organizations and interest groups may been seen as overstepping. Instead of best representing a variety of interests the public may become concerned about the undue influence that certain actors may possess in the political process. Lobbying regulation, thus, may be seen as an important way to protect the healthy competition in a competitive democracy. This chapter will seek out the causes of lobbying policy adoption in the Canadian provinces.

It is true that most democratic states experience lobbying but the structure of this differs as does the size of the interest community. While the United States is heavily described as pluralist and competitive, formalized and corporatist systems in Nordic countries allow lobbying mainly through formalized channels for workers and organized labor. Canada, while not always the first country associated with lobbying, has a large number of lobbying firms, interest groups, and consultants at both the federal and provincial level. Still, as Franks (1987) argues, interest group communities in Canada

67 differ significantly in their ability to influence and provide information from their

American counterparts. This lack of ability to influence policymakers is framed as due to the nature of Canada’s democratic institutions which stress party discipline in most parliamentary votes. It has already been stated that free votes, or unwhipped, votes where MPs may “vote their conscious” may provide an opportunity for interest group influence (Hibbing & Marsh 1987, Longley 1999, Overby & Studlar 1998, Overby &

Taydas 2011, Weale & Bara 2012). Thus, while the way interest groups and lobbyists may influence policy may differ from the classic definition of lobbying found in the

American example, interest communities in Canada exist and play a vital role in

Canadian democracy.

Until the 1990s, few regulations existed to limit lobbying or to formalize a definition of what could be considered lobbying and what lobbyists could do to influence the policy process. Since this time, regulations developed across nine Canadian provinces. These regulations often define lobbying, limit contributions to policymakers, and spell out a code of conduct for everyone from legislators to bureaucrats. Why these lobbying regulations arose is debated. It may be the case that political scandals were an important influence – spurring citizens to pressure legislators to act. In this chapter, the linkages between several important explanatory variables, including political scandal, and the adoption of lobbying policies will be tested. Using a sample which contains all

Canadian provinces from 1995-2014, this analysis will determine if scandals, which are associated with calls for change in government, do actually lead to this change.

The link between scandal and the adoption of lobbying commissions in the U.S.

States and the Canadian provinces has already been explored at some length. This

68 chapter will expand on this link by building a theory of adoption of lobbying legislation.

This analysis will then use a time-series Logit model to determine what the impact of the provinces is when political scandals occur. This chapter will utilize data collected by

Statistics Canada and also original data collected from national and provincial news sources.

By studying this connection, more information can be obtained about lobbying reforms and, more broadly, ethics legislation. While it seems counterintuitive for officials to proceed with limiting their own political power, it may be better they do it themselves than wait for citizens to become frustrated and vote policymakers out of office. If citizens are angered by political scandals, it may be necessary for politicians to vote against their own self-interest to exercise more political freedom in order to preserve their self-interest of remaining in office. While this relationship between policymakers, lobbying regulations and scandals has been established to some degree in the United

States, the comparative politics literature has yet to test this linkage empirically. This chapter focuses on the Canadian provinces. As in the U.S., Canadian provinces experience autonomy from the federal government both in terms of provincial elections and the regulation of money and influence in politics. This chapter seeks to explain why lobbying regulations are adopted in Canadian provinces, similarly to what has been done in studies of the American states.

By understanding the link between political scandal and ethics legislation, important things can be learned about the democratic process. Scandals may alert voters than there may be a need for greater regulation of government actors which may lead them to demand lobbying regulations. Lobbying policy is an area of policy where MPs

69 may first be seen as acting irrationally. Politicians have not been shown to appreciate lobbying regulations because it is thought that such laws limit the political freedoms of policymakers. However, when faced with the idea of being voted out of office, politicians and parties may choose to censor some previously observed liberties in favor of staying in power. All scandals are not equal. Certain types of scandals – those with conflicts of interest or those involving bribe taking – may have a deeper impact on the reactions that citizens may have to a given scandal (Shah et al. 2002). An index created by coding newspaper articles was created to measure the relative perception of scandals in a given province in a given year. Scandals included are required to meet the definition which views political scandals as events involving the media and political actors which debate the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. Only political scandals which meet these criteria have been included.

Theory

The mechanism examined in this chapter is that the likelihood of a province’s adoption of lobbying legislation is dependent on the number of scandals within a province at a given period of time. It is easy to assume scandals can be damaging in the minds of voters. Voters are assumed to be angry about scandals. Assuming this, voters should also want to see changes occur so that scandals become less likely in the future.

With an angry public calling for change, it is assumed that policymakers will legislate improved changes in political transparency which are favorable to them. If these

70 members of legislative assemblies (MLA’s), the term used to describe the members of provincial parliaments, act preemptively, it has been theorized that they should be able to both satisfy the public and themselves rather than worry about the backlash that could come from their party losing the next election. By acting on their own before the public is mobilized, politicians can maintain some level of control in regards to their relationships with lobbyists. This also allows MLAs to claim credit for a job done right.

A natural response both in the U.S. and Canada has been to look into lobbying reforms when political scandals occur. A political scandal is an event involving the media and political actors which debates the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. Thus, political scandals, at least as are defined here, often come down to things which appear to be “foul play” in the eyes of voters. Conflicts of interest between a regulatory body and a natural gas company or junkets to various exotic destinations on the tab of an oil company may be seen as unsavory by the public. In the past, many cases have shown that the first step to stop scandals from happening for many countries, states, and local authorities is to define what counts as “lobbying,” who is a “lobbyist,” and what is activities are allowed and what activities are prohibited under law. In Canada, the first federal lobbying regulations were adopted after the 1989 adoption of the Lobbying Registration Act in Canadian

Parliament. Among the provinces, Ontario was the first to adopt any sort of lobbying registration with the 1994 Member’s Integrity Act. However, this statute only provided information and advice on ethical questions in areas such as government contract bidding

(Office of the Integrity Commissioner 2013). Various other provinces adopted their first

71 pieces of legislation soon after –many of which created independent or semi-independent lobbying commissions in the Canadian provinces.

Subsequent updates occurred to make these lobbying commissions more powerful and also provided greater transparency. For example, most provinces now have searchable databases of registered lobbying groups and, others still, have even gone as far as creating databases that list contacts made between lobbying organizations, lobbyists, and individuals and MLA’s such as is found at the federal level or in the province of

Ontario. This requires legislators, civil servants, and members of government to document their contacts with lobbying and citizen organizations and even to declare simple “gifts” such as a cup of coffee. The regulations vary significantly from case to case. The trend of lobbying policy has continued towards more transparency over time.

Similarly, the definition of what is lobbying in some provinces has become all- encompassing. In Ontario, Section One of the Lobbyists Registration Act of 1998 (last amended in 2015) states lobbying is defined as,

” (a) in relation to a consultant lobbyist referred to in section 4 and an in- house lobbyist referred to in section 5 or 6, to communicate with a public office holder in an attempt to influence, (i) the development of any legislative proposal by the Government of Ontario or by a member of the Legislative Assembly, (ii) the introduction of any bill or resolution in the Legislative Assembly or the passage, defeat or amendment of any bill or resolution that is before the Legislative Assembly, (iii) the making or amendment of any regulation as defined in Part III (Regulations) of the Legislation Act, 2006, (iv) the development or amendment of any policy or program of the Government of Ontario or the termination of any program of the Government of Ontario, (v) a decision by the Executive Council to transfer from the Crown for consideration all or part of, or any interest in or asset of, any business, enterprise or institution that provides goods or services to the Crown or to the public,

72

(vi) a decision by the Executive Council, a committee of the Executive Council or a minister of the Crown to have the private sector instead of the Crown provide goods or services to the Crown, (vii) the awarding of any grant, contribution or other financial benefit by or on behalf of the Crown, and (b) in relation to a consultant lobbyist referred to in section 4 only, (i) to communicate with a public office holder in an attempt to influence the awarding of any contract by or on behalf of the Crown, or (ii) to arrange a meeting between a public office holder and any other person; (“exercer des pressions”) ” (Legislative Assembly of Ontario 2015; Chapter 27, Section 1).

This is a broad definition that would include anything from a large firm attempting to gain a government contract for building a new office building to a small group of mothers asking for the consideration of legislators to appropriate more funds to autism research.

Both activities would require registration with the Office of the Integrity Commissioner and violations of the act would be punished similarly.

The link which is generally supported in the literature is that when there are bigger or more-visible political scandals, there will be a push for lobbying legislation.

This link, described earlier, is likely due to the worry legislators have about losing their seats. This leads to two research hypotheses which will be tested in this chapter:

Hypothesis 4.1: Lobbying legislation will be more likely to be

adopted at times when there is the presence of political scandal.

Hypothesis 4.2: Lobbying legislation will be most likely to be

adopted when there are higher levels of political scandal.

The literature on these legislative adoptions also shows that region can play an important role in policy adoption. While scandal has theoretical links to lobbying legislation, regional influences may also play a part. Berry and Berry (1990) found that regional influences deeply influence policy adoption. The assumption is that states with

73 close geographic proximity will share common concerns, influences, and needs and, thus, emulate the legislative improvements made by neighbors. Thus, being surrounded by neighbors which have adopted lobbying regulations will increase a province’s adoption of such regulations. This leads to a possible alternative regional diffusion hypothesis that argues provinces which are abutting more neighbors will be more likely to adopt lobbying legislation than those which are surrounded by fewer geographically proximate neighbors with lobbying legislation.

The more traditional bureaucracy literature, illustrated before, dictates that countries, states, or provinces with more economic power and/or larger populations will have larger bureaucratic needs. The literature regarding bureaucratic capacity shows that heavily populated areas or areas with bustling economies require larger bureaucratic and regulatory apparatuses to ensure their government runs effectively and efficiently (Alt and Lowry 1994, Barrilleaux and Bernick 2002,Wilson 1980). Thus, as provinces become more complicated, they have required more bureaucracy. Provinces which will have stronger economic indicators and larger populations, such as Ontario or Quebec require more bureaucracy than Prince Edward Island or Newfoundland and Labrador which have much smaller economies and populations. Thus, lobbying legislation should be more likely in areas with stronger economies and larger populations due to an increased need for bureaucracy. This is another alternative hypothesis which should be explored.

Finally, political factors may further complicate this relationship. There is a link in the American politics literature between having a Democratic majority in a statehouse and the passage of lobbying or ethics legislation because Democrats favor greater

74 regulation than Republicans (Rosenson 2003). Likewise, the comparative politics literature suggests that deregulation is favored by those on the right while regulation is favored by those on the left in areas of (Benoit and Laver 2006, Boix 2000). In Canada, it is assumed that more leftist governments may be more likely to pass legislation than those who are more to the right. It is reasonable to assume that MLA’s and governments that are left-leaning will also be more favorable towards lobbying legislation. Still, this is hard to assess in the Canadian case. Because Canadian provinces each have different political parties, it would be difficult to rank each political party without something similar to a Comparative Manifesto Project (CMP) score. As CMP’s do not yet exist for provincial political parties in Canada, this is not possible. There are also no standard measures of ideology in the provinces. Thus, while it makes sense to include this in the

American case, this alternative hypothesis cannot be addressed in the Canadian provinces at this time.

Methods

This analysis will covers the time period from 1995-2014. Data are limited to this time frame due to the limitations of news coverage accessibility for content analysis.

Few records were available in certain provinces prior to 1995. Likewise, eight adoptions of lobbying regulations occurred in the time after 1995 and nine provinces adopted either their first piece of legislation or a subsequent change in legislation prior to 2014. Thus, all nine provinces that currently have some form of lobbying policy will have policy adoptions within this time period. Likewise, while some provinces made significant updates since 2014, this could lead to some censoring on both the right and left.

75

However, based on time limitations and available of news sources both before 1995 and after 2014 for the Canadian provinces, these are the best data presently available.

The statistical method used in this chapter is a time series cross-sectional analysis.

Survival analysis with repeated events is an alternative way to model the relationship between scandal and policy adoption, due to restrictions of a small sample size of only

200 observations, this was not better method to adopt. However, the results of this model, can be found in the Appendix. This time series Logit model allows comparisons to be drawn across provinces and over time. Logistic analysis is preferable due to the binary nature of the policy adoption dependent variable. Likewise, a logistical regression is able to estimate the probabilities of a lobbying policy adoption occurring and should allow for results to illustrate the likelihood that an adoption should occur.

The dependent variable used in this analysis is a dummy variable for policy adoption. This variable will be coded as “1” for an adoption of a lobbying reform in a given year and “0” as no adoption in a given year. Because it is reasonable to assume that lobbying policy will not be adopted the same year as a scandal or that lobbying policy may be adopted within the same term of a provincial parliament which spans multiple years, the change in lobbying policy occurs in a given year and the independent variables occur in the previous year. This means that if a lobbying policy is adopted in

Manitoba in 2004, the factors which should be expected to influence this should be from the year previous such as political scandal, population, etc. These factors would then be from the year before – 2003. It may take some time for the press to pick up on a political story. It may also take months of letter-writing and public pressure for the legislature to

76 respond to the public. The way these models are constructed accounts for this assumption.

Both the first passage of legislation and subsequent reforms to previous legislation will be coded as policy adoptions. Changes in lobbying policy adoptions are coded using information from two sources – the CanLii Legislative database and information published by the Office of the Commissioner of Lobbying. The Office of the

Commissioner of Lobbying is a federal agency that is charged with enforcing lobbying regulations in federal politics. On an annual basis, this agency publishes reports which show the historic overview of lobbying legislation in each of the provinces. This information lists when a lobbying commission was first adopted and also shows the latest revisions. Other revisions to legislation to laws regarding lobbying can be found on the

CanLii database, an online database of provincial legislation that shows changes in statutes across Canadian provinces over time. As discussed previously, only substantive changes in lobbying policy are considered. Subtle changes in language which occur frequently in all policy areas were not coded as subsequent adoptions of lobbying policy.

There are two measures used in this analysis which assess the key explanatory variable of scandal news reporting. Both measures originate from the survey of provincial news coverage which is previously shown in Chapter Three. The first is a binary measure of whether any political scandal was mentioned in news coverage in a given time period in a given province. Here, a province which has experienced political scandal news coverage will be coded as a “1” in a given year while a province which is free of such news coverage in that same given year is coded as a “0”. Experiencing a scandal means that a province had one or more newspaper articles coded as pertaining to

77 political scandal in the province in a given year. This addresses Hypothesis 4.2. If the research hypothesis correct, there should be an increase in policy adoptions in provinces that have experienced a political scandal in the previous year.

[Table 4.2 about here]

The second measure included in a different model is a measure of the percentage of newspaper coverage in the sample in a province which has been coded as involving a political scandal out of all articles which were found to meet the search criteria for scandal in that given year. This index is created by dividing the number of articles which were coded as mentioning political scandal by the total number of articles including scandal which were found using the search criteria in a given province in a given number of years. This is preferable to simply using the number of articles coded as political scandal because certain provinces will have many more articles than other just based on how much news coverage they have. For example, far fewer articles were found for smaller, less populous provinces such as PEI or Nova Scotia than for larger, more populous provinces like British Columbia. Thus, this index accounts for this level of proportionality. As discussed earlier, it may be the case that some provinces simply have less news coverage because of a lack of availability of certain sources over time. Due to time constraints, a collection of all news coverage in a given place in a given year would have been too labor intensive. Since the same keywords were used throughout database searches, it is assumed that this variable should be proportionate regardless of province.

An issue regarding the validity of this scandal index or percentage is that it does not account for the total share of media product that political scandals account for. A variety of other measures were proposed during the research design process including just

78 the basic number of articles coded as “political scandal”, the number of articles coded as

“political scandal” divided by the total population of a province in a given year, and the number of articles coded as “political scandal” divided by the total number of articles in newspapers from a given province in a given year. However, these measures did not prove fruitful. It was decided using an unweighted number of articles did not fully capture how much time was devoted to political scandals in a province. Total numbers of articles were hard to estimate from the search utilities of these databases. Regressions were run with a population weighted measure (where the number of coded articles was divided by the population of a province in a given year) but these results did not show any significant findings. The above scandal percentage, as previously described, seems to be an appropriate measure accounting for these limitations.

Due to differences in population across the Canadian provinces, it is expected that certain provinces are likely to have more news coverage. The mean score for a province, as seen in Table 4.1 is .475 meaning that almost half of the articles found to meet the search criteria were actually coded to meet the definition of political scandal. Meanwhile the standard deviation is .501, showing a large degree of variance exists between provinces in their degree of political scandal coverage. The maximum value for a scandal is 1.00, where all coverage from a given time met the definition of political scandal and 0 where no political scandals were coded using this news coverage.

[Table 4.1 about here]

In order to control for alternative hypotheses, three additional independent variables will be included. The first included is annual provincial population. This has been shown to increase the size of the bureaucracy overall and should be positively

79 correlated with the likelihood of policy adoption. Similarly, also included is the GDP per capita in a given province in a given year. Provinces with a larger economy are more likely to have a larger bureaucracy and should be more likely to have lobbying policy adoptions. Measures of provincial population and provincial GDP per capita come from

Statistics Canada (2016).

Finally, a measure taken from Berry and Berry (1990) will be used to assess the regional policy diffusion hypothesis. This measure is calculated for the provinces by dividing the number of provinces adjacent to a province in a given year which have previously adopted lobbying legislation divided by the total number of adjacent provinces. This creates a regional diffusion score which runs from 0 to 1 which shows the proportion of provinces which are previous adopters. Provinces with a higher regional diffusion score should be more likely to adopt lobbying regulations. It is important to note that due to the geography of the Canadian provinces, most provinces are on surrounded on one or two sides. And, to complicate matters, Prince Edward Island is truly an island that is not coded as having any “neighbors”. Thus, most provinces have low diffusion score, as seen in Table 4.1.

Results

Table 4.4 reports the results of the scandal binary model while Table 4.4 reports the results of the scandal index model. In the first model with the binary assessment of whether a province has experienced scandal, it would not appear that provinces with and without political scandal do not behave differently. The results in Table 4.4 show that provinces which have experienced political scandal coverage are not more likely to pass a lobbying policy in the following year.

80

[Table 4.3 about here]

The first model, as shown in Table 4.3, with the scandal binary does not show support for Hypothesis 1. While the estimated coefficient suggests that there could be a positive relationship between political scandal and the adoption of lobbying legislation, the association does not reach statistical significance. Based on this evidence, political scandal does not seem to have a statistically significant impact on the adoption of lobbying policies in Canadian provinces.

Regional diffusion, however, does seem to impact the passage of lobbying legislation across Canada. It appears that when there are adjacent provinces surrounding a province which have already adopted lobbying reforms, there is a higher likelihood that province will adopt legislation. This supports the regional diffusion alternative hypothesis. Over time, it does seem as though having neighbors adopt lobbying regulations will make it more likely that a given province will also adopt lobbying legislation.

Population is also linked to a higher likelihood of policy passage. This is a positive relationship that is statistically significant. Thus, as expected with the argument that more populous provinces will be more likely to adopt more regulations and have bigger bureaucracies, it seems that more populous provinces are more likely to adopt lobbying regulations. For example, Ontario was the first province to adopt lobbing regulations and is the most populous province in Canada. Thus, these results seem to support the idea that larger provinces are more likely to adopt lobbying regulations.

However, while a larger economy is also having a positive effect on the adoption of lobbying legislation in the literature, this relationship is not statistically significant in this

81 model. This is, of course, unexpected as it was assumed a stronger economy would lead to more adoptions of lobbying legislation. However, this relationship was not found to be a consistent predictor of policy adoption in the American literature. Thus, it is not a complete surprise that the economy not always lead to policy adoption. Likewise, as population and the economy may be correlated, autocorrelation may dull the effects of the economy.

In the second model, shown in Table 4.4, it is evident that the groups of provinces which had higher levels of scandal coverage or higher levels of scandal on the scandal index were more likely to also adopt legislation. While the current p-value just approaches statistical significance, it is important to note the small sample size. Over time or with a larger sample, there is a possibility that this relationship could grow stronger yet. Here, it is shown that higher scandal index values are related to a higher likelihood of the adoption of lobbying policies.

[Table 4.4 about here]

The relationship between region and policy adoption performs similarly but is a bit weaker and has a lower p-value than in the model with the nominal variable for political scandal. It still seems there is a positive relationship between regional diffusion and policy adoption in the Canadian provinces. Similarly, too, GDP per capita does not appear to have a statistically significant impact on the likelihood of policy adoption.

While the relationship presented has a similar coefficient, the economy is a less statistically significant predictor of policy adoption than in the previous model.

The fit of second model appears better. Using the scandal index may better capture scandal than the binary. This supports the argument that varying degrees of

82 political scandal may have different impacts on the public and, in turn, the legislative process. As shown in the measures of association reported, the second model detailed in table 4.4 could have a better fit. However, with such a small sample, these numbers are low. Over time, this could be improved with an increased sample size. However, this model does have a better fit than the previous judging by an improved log likelihood and better Wald Chi-Square value than the scandal binary model.

Discussion

Political scandals are a concern for political scientists, voters, and politicians.

This chapter explored the relationship between the reporting of political scandals and the adoption of lobbying legislation within the Canadian Provinces. The assumption behind the two hypotheses presented here was that scandals would increase the likelihood of the adoption of lobbying statutes. The results point to a relationship between scandal and policy adoption. However, as shown in the model with the binary measure of scandal, not all scandals can be treated equally and may not lead to policy change. It seems that the level of scandal, measured here by the amount of news coverage of scandal in a province in a given year, has an impact on policy adoptions. As scandal coverage was increased in a province, more policy adoptions are seen in this model.

In the past, the comparative politics literature shows that public pressure should lead to policy adoptions of lobbying regulations but stopped short of empirically linking these variables. This analysis has attempted to link political scandal empirically to lobbying policy adoption. Based on the results in this chapter, it is likely the case that a

83 larger number of news reports on scandal should reflect a larger presence of scandal in a province at a given time.

Thus, these results are important but are limited. They further the research on political scandal and create stronger empirical evidence that scandal does have a relationship to policy adoption in some way. This conclusion is constrained by data limitations and methodological limitations especially due to sample size but there are insights to be gained from such a study. Because the comparative politics literature on scandal is small and mainly looks at descriptive case studies, these conclusions hold weight and show encouraging evidence that scandal’s presence is not just limited to policy adoptions in specific cases or in specific national settings. Instead, this is a phenomenon that occurs in other advanced democracies.

Likewise, while Canada and the United States share important similarities – their origin as former British colonies, federalism, and similar geography – they also differ in important ways. Canadian political history and culture differs significantly from the U.S. case. Canada, a Westminster system, has different norms and has a multi-party system.

It also has stark differences between the provincial and federal party systems. The U.S. states generally mirror the federal system of parties closely. In Canada, parties at the provincial level must be considered separately from the federal level. Some systems have a great number of strongly competitive parties and who is in government may change largely from month to month. These are key differences. To see the effects that scandal has in the U.S. case translate empirically in some instances to Canada is an important finding. The Canadian provincial results broaden the types of conclusions that can be

84 made cross-nationally about lobbying legislation adoptions as these findings show these results are not only limited to the American system.

Another important conclusion which can be made with these results is the appearance of a regional effect in policy adoption. While the research on the regional diffusion hypothesis in Canada has been mixed. The overall assumption of diffusion studies has been that regional diffusion in Canada may be policy-specific. Some policy areas such as health policy show more regional diffusion than others, such as environmental policy. More research may be needed on this topic in the future but it does seem promising that region may also play into this picture of lobbying legislation provincially. Because provinces are geographically much larger than U.S. States and have fewer neighbors, the logistics of regional diffusion may be different than in the U.S.

Still, these results affirm that the adoption of lobbying policies should increase as other states surrounding a particular case also adopt these regulations. This emulation also may suggest lobbying policies may not necessarily be incredibly distinct from other areas of public policy such as environmental policy or health policies.

While this topic still requires more investigation, it is possible to see the support for the hypothesis that scandal appears to lead to a greater likelihood of policy adoption in Canada. A longer study may be required over time and may lend to, in that case, survival analysis. Likewise, a better measure of political scandal coverage may have produced stronger results. However, these results are encouraging. They show that empirical, comparative studies on lobbying policy adoption and political scandal are possible outside of case studies which focus on only one scandal. Likewise, this analysis

85 demonstrates that more fruitful research should be conducted to investigate the relationship between regional diffusion and policy adoption.

86

Table 4.1: Lobbying Policy by Province and Year

Province Year Adopted

Alberta 2007

British Columbia 2001

Manitoba 2004

New Brunswick 2014

Newfoundland and Labrador 2004

Nova Scotia 2001

Ontario 1994

Prince Edward Island NO ADOPTION

Quebec 2000

Saskatchewan 2013

87

Table 4.2: Descriptive Statistics

Descriptive Statistics N Mean Standard Minimum Maximum Deviation Policy 200 0.155 0.363 0.000 1.000 Change Scandal 200 0.475 0.501 0.000 1.000 GSP 200 40682.21 25773.17 18611.82 349350.4 Region 200 0.176 0.250 0.000 1.000 Pop 200 3211461 3777541 135737 13700000

88

Table 4.3: Scandal Binary and Policy Adoption

B Standard Error P Value Odds Ratio Scandal Binary 0.434 0.423 0.316 1.543 GDP Per Capita 0.000007 0.000006 0.282 1.000 Region 1.335 0.777 0.086 3.799 Population 0.000001 0.00000005 0.032 1.000 Constant -2.910 0.500 <.000 0.054 Rho 0.00009 Wald Chi-Square 9.860 0.043 Log Likelihood -80.985 N 200

89

Table 4.4: Scandal Percentage and Policy Adoption

B Standard Error P Value Odds Ratio Scandal Index 1.295 0.689 0.060 3.649 GDP Per Capita 0.000007 0.000006 0.304 1.000 Region 1.344 0.786 0.087 3.835 Population 1E-07 0.00000008 0.006 1.000 Constant -2.992 0.496 <.000 0.050 Rho .0000006 Wald Chi-Square 11.660 0.020 Log Likelihood -79.891 N 200

90

Chapter Five

Notes on Sponsorship: Scandal and Attitudes about Canadian Government

Introduction Chapter four explored the impact that scandal can have on the passage of lobbying legislation, but it did not explore the relationship between scandal and views of the political process more broadly. In this chapter, an analysis of public opinion data explores the relationship between respondent views of the 2004 Sponsorship Scandal and the respondent opinions about individual political efficacy, politicians, and government as a whole. The findings detail how scandal can harm the views of government and, more largely, potentially democracy as a whole.

The Sponsorship Scandal, which involved the transfer of appropriated government funding to various organizations for the purpose of Canadian promotions, meets the definition of political scandal stated herein. Various members of government abused their power and used federal funding to benefit private companies and their own pet projects. The Sponsorship Scandal is a good case for this analysis as it was a far- reaching, high-profile scandal. News about the Scandal should have reached most

Canadian voters in one way or another. The news media covered this scandal in great detail – especially from 2004 and 2006. The 2004-2006 CES cross-sectional studies and the 2006 Mail back (MBS) study allow for an exploration of public opinion on this matter with questions that ask specifically about the Sponsorship Scandal and, more generally, about views of efficacy and government.

It is important to consider how scandals affect public opinion because the literature suggests that scandals can damage individual assessments of political efficacy

91

(Villoria et al. 2007, Bowler and Karp 2004) and views of the democratic process

(Kumlin and Essaisson 2011 and Ruderman and Nevitte 2016). Democracy works best when people actively participate. A lack of political efficacy or faith in the political process can lead to greater voter apathy and abstention. Higher levels of apathy and abstention have been found to harm the health of democracies. Thus, if scandal negatively impacts the views of Canadians and reducing their desire to participate, this may harm Canadian democracy.

This chapter focuses on this important relationship between the views of voters regarding the 2004 Sponsorship Scandal and individual determinations of political ability and feelings about politics more generally. Did the level of anger about the Scandal lead to a greater disenchantment with politicians and government institutions and lower feelings of political efficacy? This chapter will illuminate how one of the most salient political events in Canadian history has affected public opinion and concludes there is a great deal of support for the idea that scandals harm individual assessments of efficacy and perhaps attitudes about government more broadly.

Theory

American and Canadian politics research suggests scandals can lead to voters to have negative views about the political process (Bowler and Karp 2004, Ruderman and

Nevitte 2004). Voters who feel less efficacious are apt to state they do not feel politically empowered or that their vote or opinion matters to politicians and government. Many things impact the level of political efficacy a voter feels – age, education, socioeconomic status, race, gender, etc. Scandal, too, has been implicated for lessening political efficacy

92 as corruption and scandal can make voters feel as though politicians are led not by political mandates from voters but due to corruption, corruption, and the whims of party elites up to no good. Last, research suggests this can lead to decrease in confidence in the federal government (Bowler and Karp 2004). This section addresses all three of these theoretical concerns and postulates three research hypotheses.

Hypothesis 1

Identifying what measure this study will use for “political efficacy” is an important step. This concept has been given numerous definitions over time. In the seminal American Voter political efficacy was seen mostly as the ability one felt they had to influence the political process (Campbell et al. 1960). In a study of European countries, Almond and Verba (1963) defined it later as “political competence” (41).

Over time, a consensus developed that political efficacy as a concept was not one specific concept but several related ones. Lane (1959) postulated that political efficacy was divided into two main concepts – external and internal political efficacy. External political efficacy is to the feeling that one has about their own individual ability to impact politics while internal political efficacy is seen as the ability to understand politics

(Converse 1972, Neimi et al. 1990, Balch 1974). Others have conceptualized a third related concept – political trust – which has, in the past, often been considered along with external efficacy. Trust is often conceptualized as the ability to have faith in the government, specific institutions, or the idea of democracy (Citrin and Muste 1999).

Primo and Milyo (2006) state, “In general, there is a consensus that trust, internal efficacy, and external efficacy are three distinct concepts” (7). The American politics

93 literature shows that these indicators of efficacy and trust, while related, are different. In the comparative literature, these concepts are often amalgamated into one particular indicator of “satisfaction with democracy”. Kumlin and Essaisson (2011) and Ruderman and Nevitte (2016) argue that scandals negatively damage this general satisfaction with democracy. However, using this general measure of satisfaction with democracy may not fit every scenario or research question. While it is justified in the case of Ruderman and

Nevitte to include “satisfaction” due to their research question about the concerns of faith in democracy after the Sponsorship scandal, it would not directly shed light on a research question about specific views of government and individual political indicators.

For example, Primo and Miylo (2006) study the effect of campaign spending limits in American states on political efficacy. They justify the use of internal and external efficacy indicators because they were concerned with how these specific laws impacted the abilities of voters to either understand politics or participate in politics. A measure of government trust is not preferable when more specific indicators were available. Bowler and Karp (2004) also employ a similar approach to studying the individual impact of political scandal in a study of the UK and US, using an external political efficacy measure. The measure used in this study, which will be discussed later and is included in the appendix, is quite similar to the question employed in both Primo and Milyo (2006)’s study and Bowler and Karp (2004)’s study. This measure originates in the CES. This indicator of political efficacy more reasonably measures the ways in which scandal may impact assessments of individual political indicators.

In the case of the Sponsorship Scandal, incredibly public allegations of political corruption and widespread violations of electoral law appear to have damaged the

94 credibility of the Liberal Government. However, voters may feel that this scandal also damaged their ability to guide political decision-making. Disheartened, voters may have felt, in the time following the Sponsorship Scandal, that they were rather powerless. In particular, the level of anger expressed towards the scandal should be related, inversely, to external political efficacy. Angry voters are less likely to feel particularly high levels of political power. Hypothesis 5.1 addresses this concern:

Hypothesis 5.1: The more anger a respondent feels towards the

Sponsorship Scandal, the less external political efficacy the respondent

will feel.

Hypothesis 5.2

Bowler and Karp (2004) suggest that the results of scandal range beyond general views of political efficacy. Their findings suggest people who were affected by scandal should have more negative views of particular actors and institutions. The Sponsorship

Scandal involved high-ranking party leaders and federal politicians including the Prime

Minister Chrétien. This Scandal not only damaged reputations of individual party members and politicians, it also saw the end of long-held Liberal dominance of the

Canadian political system. Thus, it is reasonable to assume it may have impacted a large number of Canadians’ views about more than just personal political efficacy but also views of the federal government and politicians in Canada.

For this reason, this research posits another theory. It assumes voters may feel negative about politics in general. Citizens may begin to blame all politicians for the sins of a few. Thus, while those implicated were generally political elites in the Liberal Party,

95 voters may not differentiate what party was to blame. Instead citizens may have felt as though all politicians are involved in scandalous acts. In the Canadian case, while the

Liberal Party was implicated in the scandal and Jean Chrétien, the Prime Minister, was found to be involved, not all Liberal MPs were implicated. For example, Paul Martin, who became the eventual party leader of the Liberals after Chretien, was ultimately absolved of any responsibility. Still, Canadian voters may view all politicians with a certain degree of suspicion after such a scandal. The surfeit of evidence on the relationship between scandal and political behavior suggests overall approval of politicians and government declines following a political scandal. Thus, it should be relatively safe to assume Canadians in 2006 view politicians with a certain degree of distrust. To address the concerns of views of politicians, Hypothesis 5.2 is posited:

Hypothesis 5.2: The more anger a respondent feels towards the

Sponsorship Scandal, the more negatively a respondent will feel

about politicians.

Hypothesis 5.3

Finally, this remaining distrust should go beyond trust in politicians and should affect views of the federal government as a whole. Similar to the assumption that acts of a few politicians can dishearten voters and change voter opinions of all politicians, the literature has found that a large-scale scandal can deeply impact the views of voters towards large institutions like Parliament or Congress (Bowler and Karp, 2004).

Moreover, while certain other political actors such as MPs and members of government agencies were found to be involved, few people involved in daily Canadian Government

96 were found to be participants in the Sponsorship Scandal. Voters, however, may not distinguish between the findings of the Gomery Commission and the press and the reality they believe to be true. For this reason, Canadian voters may feel a degree of distrust in the federal government as a whole during this period of time. This theory about the final hypotheses:

Hypothesis 5.3: The more anger respondent shows towards the

Sponsorship Scandal, the less confidence in the Federal

Government a respondent will feel.

Methods

This chapter will use three OLS regression models. The data utilized come from the 2006 wave of the CES to determine what the relationship is between views of political scandal and views of political efficacy, politicians, and government. This will require the use of several regressions to analyze the impact of anger over the Sponsorship

Scandal on political opinion. The first two regressions will use data from the 2006 CES cross-sectional study. The 2006 wave of the CES included a number of questions directly addressing the Sponsorship Scandal. The third regression will requires a variable pulled from the 2006 MBS, which has a smaller sample size and asks questions about different political issues facing voters than are included in the 2006 CES. The MBS, as was described in Chapter Three is a mail back survey that is self-reported rather than administered by an interviewer. Because questions about government indicators such as confidence in the federal government were not asked in the 2006 CES cross-section, the

97

MBS has been utilized. More detailed information on all of these variables is available in the Appendix, which contains full question wording and responses.

The first dependent variable used is political efficacy. This measure originates in

2006 CES and asks respondents to identify how strongly they agree or disagree with the statement, “I don't think the government cares much about what people like me think.”

This is a good proxy for external efficacy because it measures on a scale from 1-4 how much power a respondent feels they exercise over government. This is a fairly classic

“NOSAY” question, as is seen in the American political behavior literature (Craig et al.

1990). Craig argues that NOSAY questions address why voters feel they have “no say” in political affairs. Craig argues that these questions tap external political efficacy most effectively. On this scale, a value of 1 indicates that a voter strongly agrees and a value of 4 indicates that a voter strongly disagrees with this statement. Higher numbered responses indicate higher levels of voter efficacy. The average respondent indicates they have some degree of political efficacy and is in between 2 and 3 on this scale, as can be seen in Table 5.1 which lists descriptive statistics. The standard deviation is about 1, indicating that few respondents list themselves as being very efficacious.

The second dependent variable included is the respondent’s view towards politicians. The measure taken from the 2006 CES asks the respondent “how do you feel about POLITICIANS IN GENERAL?” Responses adhere to a 0-100 thermometer scale where “0” indicates that they do not feel warmly at all to politicians and “100” indicates the highest feelings of warmth towards politicians. This allows the voter to determine how, in general, they feel about politicians. Higher numbers equate to higher levels of warm feelings towards politicians. In Table 5.1, the average score given by respondents

98 for feelings towards politicians is near the middle at about 44 points. The standard deviation indicates that people, in general, are not likely to rate politicians either very coldly or incredibly warmly.

[Table 5.1 about here]

Finally, for the third model, the dependent variable is trust in government. The measure assesses trust in government comes from the 2006 CES MBS and asks a respondent, “How much confidence do you have… in the federal government?” The variable is coded on a 4-point scale from “1” which means “none at all” to “4” which means “a great deal.” While this measure does not specifically inquire to the level of trust one has in the government, confidence as a concept is closely related. Also, while the question is vague, one can assume that “federal government” is a fairly wide concept which covers the both the narrow definition of those in government and the larger view of “government” which includes all of the legislative branch and bureaucracy. This allows for a more holistic assessment of how the respondent views the Canadian government. Lower numbered responses to this question will indicate a lower level of confidence in government. The mean for confidence in government, as shown in Table

5.1, is between 2 and 3 and indicates that most respondents have some level of confidence in government. The standard deviation suggests that this rating varies but that most respondents stay between a range of 2 and 3. Few respondents either have no confidence in government or a very high level of confidence in government.

One problem with using this measure comes from a reduction in sample size.

This question was only asked in the 2006 MBS, which allowed participants in the 2006 cross-sectional study to mail in responses to additional questions. This may lead to a

99 certain degree of self-selection bias in this question. While the cross-sectional study was conducted through interviews, this survey was administered differently. Thus, the respondents which decided to participate in this survey may have different opinions than the sample of the CES cross-sectional study.

The primary explanatory variable is the level of anger a respondent expresses towards the Sponsorship Scandal. This measure originates in the 2006 cross-sectional

CES. The question asked in the survey is, “Now some questions about the Sponsorship

Scandal. Does it make you very angry, somewhat angry, not very angry, or not angry at all?” The responses have been coded from 1-4 where “1” equals “not very angry at all” and “4” equals “very angry.” Higher numbered values indicate higher levels of anger.

The mean of this variable, as shown in Table 5.1, is over 3 points, indicating that a large number of respondents are angry about the Sponsorship Scandal. The standard deviation suggests that most respondents fall between a moderate level of anger and a high level of anger towards the Scandal but that few respondents are “not very angry at all” about the

Sponsorship Scandal.

Also included in all three models are control variables which are included to rule out alternative hypothesis and intervening relationships. All control variables included come from the 2006 CES cross-sectional study. More detailed explanations of these control variables are also included in the appendix. Because being involved in politics is correlated with higher levels of political efficacy and greater feelings of political trust, two dummy variables are included which assess a respondent’s participation in community volunteer organizations and political organizations. As news readership may be correlated with higher efficacy and confidence in government, a measure assessing

100 how much one read in the newspaper about the federal election is also included. Also, since education is linked to both better political sophistication and higher levels of wealth, a measure was created to assess the level of education of the respondent.

Age also may influence political efficacy and trust in government, so it is included as well. However, as many political behavior studies have shown, age may actually have a cyclical relationship to these variables. For this reason, age squared has been included. Finally, because those who express allegiance to the Liberal Party may potentially be less likely to express anger about the Sponsorship Scandal overall, a control variable is included to assess partisanship. Because of the nature of the Canadian political system and the lack of questions about ideology which are included in the CES, a measure was crafted by subtracting a thermometer scale which assesses one’s feelings towards the Conservative from the thermometer scale which assesses one’s feelings towards the Liberal Party. The difference between these scales can range from -100 to

100. Those who have higher numbers on this scale have more warm feelings to the

Liberal Party and may be less angry over all and may feel higher levels of political efficacy and trust in government and politicians and they may be less impacted by the

Sponsorship Scandal.

OLS regressions are used to determine the effects of the Sponsorship Scandal on political efficacy, views of politicians, and trust in government. While none of these variables use continuous measures, the enhanced ability to generate straightforward results outweighs the benefits of using an ordinal Logit model. OLS models allow for the ability to determine size of the impact in a much more straightforward manner than an ordinal Logit model. There are drawbacks to this approach, of course. Because there are

101 not equidistant differences between one response and the next, there is no direct which can be made between these coefficients and those of continuous variables. It should be noted, however, that the interpretation of these results should not lead to assumptions of proportionality of changes between variables. For example, a response that is a “3” on a scale of a political efficacy is not only 1 proportional step from a “4”. However, with two variables with a 4-point rating scale and one variable with 101-point rating scale, there should be enough categories for a reasonably reliable set of results.

Results

Hypothesis 5.1

Table 5.2 displays the results for the first regression are included. This table describes the relationship between anger over the Sponsorship Scandal and the feelings of political efficacy respondents reported. This statistical model tests Hypothesis 5.1 which postulates that as feelings of anger over the Sponsorship Scandal increase, feelings of external political efficacy should decrease. The results of Table 5.2 suggest that this relationship holds. There is a strong, negative relationship between the way a respondent feels about the scandal and the way they evaluate external political efficacy. An increase in one level of anger towards the Scandal reduces the reported level of external efficacy in a respondent by one level. When respondents are angry with about the Sponsorship

Scandal, they feel as though the government does not care about their opinions on political matters. Anger with the Sponsorship Scandal seems to translate into voters feeling less powerful politically. Overall, this strong coefficient shows the strong effect of the Sponsorship Scandal on political efficacy.

102

[Table 5.2 about here]

Other indicators which increase political efficacy are being a political volunteer or feeling warmly towards the Liberal Party. These results are expected because more politically empowered people are likely to feel more efficacious and those who are sympathetic to the Liberal Party are probably less angry about the Sponsorship Scandal and more likely to report higher levels of political efficacy. Paying attention to the news and being more educated are often correlated with higher levels of political efficacy due to their presence in more politically sophisticated voters. These assumptions are supported in this regression. Age’s relationship with efficacy is weak and only marginally statistically significantly related to feeling less efficacious. Finally, reporting a respondent was a community volunteer does not appear to have an effect on the level of political efficacy one reports. This is not as expected but it seems as though volunteering in one’s community does not make a respondent feel any more positively about whether the government cares what they have to say.

Hypothesis 5.2

The results for Hypothesis 5.2 are displayed in Table 5.3. Here, as expected, the table suggests that voters who have more anger towards the Scandal are more likely to feel less warmly about politicians. This is a negative, statistically significant relationship where a 1 point increase in anger about the Sponsorship Scandal leads to about a 1.5 point decrease in feelings about politicians. This is not a large decrease but it is still a significant increase which indicates that a respondents who feel more anger over the

103

Sponsorship Scandal seem to feel more negatively about politicians. Hypothesis 5.2 is supported by these results.

[Table 5.3 about here]

Other variables in this model are statistically significant. Voters who are older appear to have a more negative view of politicians. However, these variables only seem to make respondents view politicians more positively. As expected, Liberal-minded view politicians more warmly because they are likely to feel less angry about the Sponsorship

Scandal. Volunteering, as well, also seems to make people feel more warmly towards politicians – both for a political campaign and for community organizations. Finally, newspaper readership and education make respondents more positively view politicians.

These results mirror the results for the first model for political efficacy.

Based on the above results, there was support found for Hypothesis 2. Those who are most angry about the Sponsorship scandal are most likely to feel most negatively towards politicians as a whole. While most other indicators of partisanship and political sophistication led to warmer feelings towards politicians, intense feelings about the

Sponsorship Scandal were predictive of more negative feelings towards politicians.

Hypothesis 5.3

While it appears that there is a relationship between views of politicians and views of political scandal in Canadian voters, the larger issue of how scandal may impact a respondent’s view of government has yet to be determined. Results presented in Table

5.4 show strong support for Hypothesis 5.3. Here, an increase in anger over the

Sponsorship Scandal bring leads a respondent to have less trust in government. These

104 results are statistically significant and in the expected direction. A one point increase in anger reduces confidence in government by .135. Thus, scandal may not only harm the views of politicians broadly, it can damage the overall credibility in government institutions.

[Table 5.4]

Similarly to the findings above, age also seems to lessen confidence in government. Older respondents are more likely to indicate that they have less confidence in the federal government. Other variables seem to improve confidence in government.

For example, being involved in political campaigns seems to increase confidence in government. This seems logical as those more involved in politics should view political institutions more positively. Finally, being more politically warm towards the Liberal

Party is positively associated with a more positive view of government. Again, these respondents likely view government more positively due to their support of the scandal- embroiled Liberals. More educated voters also tended to view the government more positively.

Feelings about the Sponsorship Scandal and the view of government seem strongly and significantly linked based on the results and seem to bolster strong support for Hypothesis 3. It appears that scandal not only has a negative impact on the views of politicians who are often seen as mired in scandal but also attitudes towards the federal government as a whole. Overall, though, the results in Tables 5.2 through 5.4 support all three hypotheses. Not only do views of scandal negatively impact one’s feelings of political efficacy, they may make a respondent feel particularly negative about political

105 actors and the political process as a whole.

Discussion

This chapter explains the relationship between views of scandal and public opinion. Examining the case of the 2004 Sponsorship Scandal has allows for a more nuanced view of how voters react to political scandal more broadly. This case study also aligns closely with the statistical analysis in the American Politics literature. The results included in this chapter suggest that scandal can have broad and discernible impacts not only on personal feelings about the ability to harness political power but, also, in the way a person views government. In all three models, it was shown that a negative view of the

Sponsorship Scandal hurts assessments of government– either through views of actors or through the way one perceived their role in government.

This can lead to negative implications for democracy in general, of course.

Voters who do not feel empowered and who have limited trust in government tend to abstain from politics. This can weaken the entire democratic process. What may have been a democratic mandate at one time is now weakened due to low turnout even for national elections. While these results only tap into the views of Canadian voters, it could be argued that this trend likely exists elsewhere.

It is unknown how long the effects of the Sponsorship Scandal lingered. A

Liberal victory in the 2015 elections may suggest that these effects have lessened. This is why it would be, perhaps, more helpful to look at these results over a long period of time.

While the CES does keep a panel, which will be examined in greater detail in Chapter

Six, the time trend with this panel is short. Thus, it is difficult to say what the results of

106 these models would be three, five, or ten years from the date of the scandal. There is also no comparison that can be made between this scandal and another type of scandal.

The questionnaire used focuses primarily on general political indicators and selective topics – one of which is the Sponsorship Scandal. In order to make some conclusions with greater external validity and to learn more about scandal over time, a long time series would be helpful. Also, perhaps it would be useful in the future to compare the

Canadian case to other similar cross-national cases of scandal. This would allow for more variation in the sample. Chapter Six will attempt to look at changes in public opinion between 2004 and 2008 and will expand this discussion to include provincial scandals. By tying in the provincial scandal data used in Chapter Four, it will be possible to see variations in public opinion between respondents who are in provinces which have experienced political scandal and those respondents who have not experienced recent political scandal. While not a comparison across countries, this analysis allow for a comparison across varying provincial cases with the same electoral and political system.

However, looking only at the results in this chapter, it is possible to draw some broader conclusions regardless of the lack of time trend or cross-national sample. The first and perhaps most important point shown in this analysis is that scandal matters to voters and has an impact on public opinion. The results of this impact can be seen most basically in the very large historic losses by the Liberal Party. The Liberals were punished harshly for the Sponsorship Scandal and only recently in the 2015 elections came back to power under Leader and Prime Minister Justin Trudeau. The Sponsorship

Scandal had serious implications for politics in Canada. The results in this chapter show just how damaging and wide-ranging the effects of political scandal can be.

107

Table 5.1: Descriptive Statistics

N Mean Standard Minimum Maximum Deviation Efficacy 3075 2.245 0.959 1 4 Confidence 1628 2.745 .730 1 4 Politician 4126 41.774 23.24 0 100 Angry 3981 3.19 0.848 1 4 Political Volunteer 1684 0.17 0.376 0 1 Community 1684 0.72 0.449 0 1 Volunteer News 4299 3.445 3.252 0 10 Education 4267 1.853 0.811 1 3 Age Squared 1992 2987.088 1779.465 400 8464 Liberal- 3820 -696 33.621 -100 100 Conservative Thermometer

108

Table 5.2: Anger and Political Efficacy

Coefficient Standard Deviation P-Value Angry About -.932 0.030 0.002 Sponsorship Scandal

Political Volunteer .134 0.062 0.029 Community Volunteer 0.069 0.055 0.201 News Attention 0.015 0.007 0.031 Education 0.123 0.03 < .000 Age Squared -0.00002 0.00001 0.1-0 Liberal-Conservative 0.094 0.001 < .000 Thermometer

Constant 1.805 0.135 < .000 F Statistic 29.59 < .000

R Squared 0.121 N 1517

109

Table 5.3: Anger and Views of Politicians

Coefficient Standard Deviation P-Value Angry About -1.618 0.632 0.011 Sponsorship Scandal Political Volunteer 7.199 1.288 < .000 Community Volunteer 1.937 1.133 0.087 News Attention 0.655 0.151 < .000 Education 2.127 0.626 0.001 Age Squared -0.006 0.0003 0.059 Liberal-Conservative 0.407 0.02 < .000 Thermometer Constant 21.85 2.809 < .000 F Statistic 93.59 < .000 R Squared 0.304 N 1507

110

Table 5.4: Anger and Confidence in Government

Coefficient Standard P-Value Deviation Angry About -.135 .026 <0.000 Sponsorship Scandal Political -.015 .051 .773 Volunteer Community 0.015 .047 .034 Volunteer News Attention .004 .006 .519 Education .09 .025 0.001 Age Squared .00006 .00001 0.059 Liberal- . 2 .0008 < .000 Conservative Thermometer Constant 3.324 .121 < .000 F Statistic 56.78 < .000 R Squared 0.287 N 972

111

Chapter Six

Provincial Scandal and Public Opinion: A Look at the 2004-2008 Panel Data

Introduction

In Chapter Five, data from the 2006 Canadian Election Study was used to show the ways in which political scandal can be damaging to the view of voters about the electoral process. In this chapter the 2004-2006-2008 CES panel data moves this argument forward by examining if views of provincial scandal have an impact on the views of the federal government and general political efficacy. This requires a pairing of the CES data with the original scandal data described in Chapter Three. By merging the information gathered from Canadian news sources and a set of panel data collected by the

CES, this analysis illuminates how scandals color voters opinions of personal political efficacy and on the larger political process.

In this chapter, the opinions of Canadians will be examined on two topics: personal political efficacy and views towards politicians in general. Using an OLS regression analysis, this chapter examines if those respondents who are from provinces which have experienced a political scandal recently will experience lower levels of political efficacy and less favorable views of politicians. Likewise, panel data allow for a view of differences of opinions in voters over time. Thus, unlike Chapter 5 where questions about scandal and its effects on public opinion were only available at one point in time, this survey gives three data points by which to see changes in public opinion.

112

Theory

The focus in this chapter is on public opinion and political scandal; the awareness of political scandal negatively impacts political efficacy and evaluations of politicians.

News coverage of political scandal is expected to lead Canadians to view the political process in a less positive light than those who live in a provinces without a recent political scandal. Thus, the most basic prediction is that respondents who have experienced a recent provincial-level scandal will view politics in a more negative light than those who have not experienced a provincial-level scandal.

Still more recent public opinion research which focuses on federal countries, shows that local or sub national-level scandals may impact citizen evaluations of politics and government not only at the local and sub national levels of government but also at the federal or national levels of government (Gélineau and Remmer 2005; Ulbig and Miller

2012). This is due in part to the inability of voters to determine what levels of government are responsible for which programs or, in this case, which level of government has experienced recent indiscretions (Cutler 2004, 2008; Whitten and Powell

1993). It can be difficult for a voter to determine whether the true cause for concern is a federal political party that took a bribe or a provincial party or both. This ambiguity may lead to all politicians and government actors being cast in a negative light. Thus, there may be a reason to believe that views of political scandal may “bleed” from one level of government to another.

Effects of scandal on public opinion may be quite limited, however. As shown in the previous literature review, the effects of scandal over time dwindle as people often forget scandals once a few years have passed or at least view them with less concern over

113 time (Praino et al. 2013, Kumlin and Essaisson 2011). For this reason, the scandals included in this study will be limited to the year before and the year of the data point. For example, for the survey in 2004, only scandals in 2003 and 2004 would be included for a given province. Only recent scandals will be included in this analysis.

Hypothesis 6.1

Political efficacy has been shown to be negatively correlated with anger towards scandal (Bowler and Karp 2004). This relationship is supported, as well, in the results from Chapter Five. Here, it was found that those respondents expressing more anger towards the Sponsorship Scandal were likely to feel lower levels of political efficacy.

There is an assumption that voters who endured news coverage of provincial scandals will have distinct or different political opinions from those voters who live in scandal- free provinces. A respondent from an area fraught with recent scandal will feel less politically efficacious than someone who is from an area which has not recently experienced a political scandal. This leads to the first hypothesis of this chapter:

Hypothesis 6.1: A respondent from a scandal-ridden province will feel less

political efficacy over the time after a recent scandal than a respondent

who is from a scandal-free province.

Hypothesis 6.2

Views of more general parts of the political process are also expected to be affected by political scandal. This includes, perhaps most obviously, the view one holds for politicians. Even if a respondent does not particularly dislike their political

114 representative, it has been found that political scandals have been linked to poor evaluations of all policymakers. And, as seen in Great Britain, a voter may not even be particularly aware of whether their particular representative was involved in a political scandal but will still be apt to view all politicians with suspicion (Bowler and Karp 2004).

This conclusion was supported as well in the findings of Chapter Five, where respondents who expressed more anger over the Sponsorship Scandal were more likely to feel less warmly about politicians, in general. Moreover, this feeling may from one level of government to another, as was seen in the more recent literature on public opinion, scandal, and federalism. When asked a general question about politicians or political institutions, it should be possible for scandals at the provincial level to bleed into the federal level or into more general views of politics. This leads to a second hypothesis:

Hypothesis 6.2: A respondent from a scandal-ridden province will be

more likely to view all politicians negatively over time than will a

respondent from a scandal-free province.

While the CES Panel looks mainly at federal politics and policies, it allows for comparisons to be made across groups and, potentially, across levels of government. Not all provinces in this sample experienced scandal in the time briefly before each of these surveys was taken. Thus, there are variable levels of scandal allowing for the comparison between those who come for provinces with scandal and those who come from provinces without scandal. As stated before, while we would expect that provincial scandals would have the greatest impact on opinions about provincial politics, recent literature suggests that these feelings can bleed over into more general conceptions of politics and even

115 federal politics. Thus, the 2004-2008 CES does enable a test of all of the above hypotheses.

Methods

Several OLS models will test two research hypotheses. Three separate dependent variables will be used throughout this analysis. While the 2004-2006-2008 panel allows for three data points, time series cross-sectional estimation is not appropriate. Similar to many economic voting models which use data over only a few time points, a change model allows for a regression of key independent variables on a dependent variable which takes the difference of a score in one time period and the score from another time period. These three analyses divides these scores into three time periods allowing for the change between dependent variables in 2006 and 2004, the change between dependent variables in 2008 and 2006, and the change between dependent variables in 2008 and

004. This will allow for some appreciation of the change in public opinion between 2004 and 2008 across all 10 Canadian provinces.

Dependent Variables

The dependent variables originate in the 2004-2006- 2008 CES panel surveys.

Each dependent variable is based on a question that was asked during the 2004, 2006, and

2008 waves of the sample. These variables will be described in some detail here but more information is available in the appendix covering the variables from this chapter.

Each dependent variable is a change variable meaning that one year’s score will be subtracted from the following year’s score to create a difference variable. Each

116 dependent variable used has three scores – a 2008-2006 score, a 2006-2004 score, and a

2008-2004 score. The panel respondents who dropped from the panel or who fail to give responses to key questions were not included in the analysis. This accounts for some of the differences in sample size seen in Table 6.1, which shows descriptive statistics for variables in these models. Thus, because of this panel there are differences in response rates across questions and time periods.

[Table 6.1 about here]

The first dependent variable tests for political efficacy in Hypothesis 6.1. The corresponding question from the 2004-2008 CES Panel asks the respondent to state their level of agreement with, “I don't think the government cares much about what people like me think.” The respondent then states on a 4-point scale how strongly they agree or disagree where 1 is “strongly agree” and 4 is “strongly disagree”. This is the same question which was used in Chapter Five to assess political efficacy.

The change variable for efficacy, however, differs from the measure used in

Chapter Five as does the interpretation of the variable. As efficacy increases over time, it is expected scores will get larger. If there has been a decrease in political efficacy over time, scores will get smaller and. For example, if in 2004 a respondent reported a rating of “4”, which shows a high degree of efficacy, and then reported a score of “2” in 2008, in the 2008-2004 change model, this would give the respondent a rating of “-2” which would show an overall decrease in political efficacy over time. In Table 6.1, the mean values for the three change dependent variables indicate that the average respondent felt less efficacious in 2006 than 2004 and less efficacious in 2008 than 2004, as expected.

After all, it is assumed that at the height of the expense scandal voters would feel more

117 negatively about politics. However, between 2006 and 2008, efficacy increases slightly.

This relationship is also displayed in 6.1 where it appears in all provinces, there is a drop in political efficacy between 2004 and 2006 and an increase in efficacy again in 2008. If

Hypothesis 6.1 is supported, it is expected that respondents from scandal-ridden provinces will experience reduced feelings of political efficacy in comparison to those respondents from scandal-free provinces.

[Figure 6.1 about here]

Another dependent variable tests Hypothesis 6.2 which argues that those respondents who experience scandal at the provincial level should have more negative opinions of politicians. The question used to test this hypothesis originates in the 2004-

2006-2008 CES Panel. Here, respondents are required to assess how warmly they view politicians as a group from 0 to 100 where 0 corresponds with a great dislike of politicians and 100 corresponds with a great like of politicians. This is the same question used to measure feelings of politicians in Chapter Five.

[Figure 6.2 about here]

The politician thermometer change variable does change the interpretation from the politicians thermometer used in Chapter Five, however. It is expected that a respondent who feels less warmly about politicians should rate them lower over time.

This should lead to larger, positive values being associated with respondents who feel more negatively about politicians over time and smaller, negative values being associated with respondents who feel more positively about politicians over time. For example, a respondent who responds with a score of “60” in 2004 but then a score of “20” in 2008 would have a score of “40”. Meanwhile, a respondent who responds with a score of “20”

118 in 2004 and “60” in 2008 would have a score of -40. Looking at table 6.1, the average respondent’s change score is in the positive 20s across all time periods indicating that they likely view politicians less positively over time. The visual change in the mean response to this question in 2004, 2006, and 2008 is available in Figure 6.2. While attitudes towards politicians recovered a bit in 2006, most respondents viewed politicians less positively in 2004 and 2008. Once more, the assumption with this variable is that those who are from scandal-ridden provinces will feel lower-levels of trust in politicians and will feel less warmly about them.

Independent Variables

The main explanatory variable included in these models measures political scandal using original data collected from a content analysis of news coverage of scandals in the Canadian provinces. This is the same scandal binary used in Chapter

Three. A political scandal here is defined as an event involving the media and political actors which debates the wrongfulness of an alleged political transgression regarding a conflict of interest, money in politics, or using one’s office for financial gain rather than to the benefit of constituents. A “recent” scandal is one which meets the definition of political scandal and has occurred either in the year the survey was conducted or the year of the survey. Thus, to be coded as a “1” during the 2008 survey period, a respondent would have to have been from a province in which a scandal occurred between 2007 and

2008. Those respondents who are from provinces coded as “0” are from provinces which did not recently experience a political scandal while those who have been coded as a “1” are from a province which experienced a recent political scandal. These numbers are

119 pulled from the Scandal Data described in Chapter Two and which are used in the analysis conducted in Chapter Three. Those respondents who are from provinces which have recently experienced scandal should be more likely to feel negatively about politicians and have lower political efficacy. An explanation of how provinces are coded is shown in Table 6.2

[Table 6.2 about here]

The models will also include a number of control variables which account for other factors that could influence public opinion. These variables originate in the 2004-

2006-2008 CES Panel and more information about them and how they are coded can be found in the Appendix. First, this analysis will use a square for age. Age has been shown that public opinion may act cyclically with age. While younger voters may lack efficacy, so, too, may the oldest of voters. Different age groups may also view politicians more negatively and may also feel different levels of political efficacy. More educated voters may also differ from less-educated voters in that a higher level of education tends to lead to greater personal wealth. Education has been found to be an important determinant of political efficacy and feelings about government. As such, this will need to be included. Because education is highly correlated with wealth, another oft-cited determinant political sophistication and engagement, income was not included.

Education was used as a proxy for socioeconomic status as people are more likely to know their educational history than their gross income

A measure of news coverage will also be included. This comes from a question which asks how frequently a respondent has paid attention to federal elections in the newspaper. Respondents who are more aware may feel more efficacious but also may be

120 more aware of scandals and, thus, more apt to blame politicians for bad behavior. As there is no measure for provincial news readership or general news readership which was included in each of the CES surveys, this proxies for readership in general. Obviously, a measure which included coverage of provincial readership would be preferable but this does not exist in these surveys.

Finally, because these scandals do not exist in a vacuum, a measure of federal partisan identification must also be included. As Wagner et al. (2014) discovered, partisanship can have a strong impact on the ways voters recall scandals. People who are partisan will be most likely to ignore when their party missteps. Political views may be especially important in the time surrounding the Sponsorship Scandal because the Liberal

Party was identifiably involved in the political scandal. Thus, a thermometer score was created to show how warmly a respondent felt towards the Liberal Party in comparison to the Conservative Party. This was done similarly to the measure used in Chapter Five, where the Liberal Party thermometer score was subtracted from the Conservative Party thermometer score. This measure is not perfect and may not correlate with opinions towards political parties at the provincial level. However, there are no questions which ask about party identification provincially and creating a left-right scale with all provincial parties would be impractical in the case of Canada. Thus, this political thermometer score will be used to assess Liberal identification. Higher scores identify stronger Liberal support while lower scores identify a stronger tie to the Conservative

Party.

121

Results

Hypothesis 6.1: Scandal and Efficacy

The results in Table 6.3 do not lend support to Hypothesis 6.1. It was expected that there would be a negative association between political efficacy and scandal. People from scandal-ridden provinces should report lower levels of political efficacy. Changes reflecting lower levels of political efficacy with the dependent change variable in these models are only seen in one model. The 2008-2006 change model shows a relationship in the expected direction but the relationship is not statistically significant. In the 2006-

2004 model, there is a significant relationship between being from a province which has recently experience scandal and a change in political efficacy. However, this change shows was in the opposite direction than predicted. In this case, being from a province that had recently experienced scandal increased political efficacy ratings by about a tenth of a point. Over the time of the entire panel from 2004 to 2008, similar results are also reported. Here, being from a province which was coded as having a recent scandal raises reported values of political efficacy over time by about a tenth of a point. This seems to suggest that federal provincial scandals do not make voters feel more or less efficacious overall.

[Table 6.3 about here]

When it comes to federal politics, it appears that other variables are likely to have a deeper effect on political efficacy for Canadian respondents. Age was found to negatively impact views of political efficacy but the effect was very small and only found in the 2006-2004 model. News readership was only associated with a significant decrease in efficacy in the 2008-2006 model. The only variable with a statistically

122 significant relationship across all three time points is the Liberal-Conservative thermometer measure. However, the direction of the coefficient changes throughout, meaning that during the 2008-2006 and 2008-2014 sample, partisanship may have decreased political efficacy while in 2006-2004 it may have increased political efficacy.

Overall, the fit of these models is poor. The R-squared values are low, especially in the 2006-2004 model. The F statistic in this model as well, while significant, is incredibly low. The 2008-2006 and 2008-2004 models perform better than the first model but none of these statistical tests performed especially well. It would appear, thus, that federal politics, while perhaps not thought of completely separately from provincial politics, are not always linked in the minds of voters. While Canadians may be harmed by provincial scandals in some ways, these results do not support that provincial scandals impact views of political efficacy federally.

Hypothesis 6.2: Scandal and Views of Politicians

In the models which test the impact of provincial scandal on views of federal politicians, there does not seem to be strong support for Hypothesis 6.2. All coefficients reported detail a strong, positive association between a respondent’s province having a political scandal and the change in the politicians thermometer. This was expected. As views about politicians decrease over time, these scores rise based on the way this index was calculated. Thus, there should be a positive relationship between the scandal binary and the change in the politician thermometer. Still, this relationship is only statistically significant in the 2008-2004 model. This result suggests that a respondent from a scandal-ridden province should see a two-point decline in the politicians thermometer

123 score between 2004 and 2008. Thus, there may be some limited support for the idea that, even at the federal level, politicians are viewed more negatively by those respondents from provinces who have recently experienced a political scandal. However, this relationship is not very statistically significant and is only supported in one of the models below.

[Table 6.4 about here]

Other variables showed some predictive power in the change in the politician thermometer. In two models, the age variable was associated with a change in the politicians thermometer but only weakly so. Age negatively impacted views of politicians in the 2006-2004 mode and positively impacted politicians in the 2008-2006 model, which is inconsistent. The 2008-2004 model showed no statistically significant relationship between age and views of politicians. Education was associated with better views of politicians in only one time period (2006-2004). Partisanship, in this set of models, performed as postulated. Across all models, there was a negative association between views of politicians and being more partisan towards the Liberals. Here, respondents who viewed the Liberals more warmly were apt to rate politicians more warmly between about a third and three quarters of a point on the political thermometer.

Partisanship, as was shown in Chapter Five may make it easier for some voters to recover from scandal than others.

Again, these models appear poorly fitted. The R-Square values for each model are very low. Likewise, the F Statistics reported are fairly small but all are highly statistically significant. However, the F Statistic only describes that there is some association between variables. It cannot show strong support for the hypotheses tested.

124

The politician thermometer models may fit slightly better but they still lend support to the previous literature which believes that members of federal states can identify clarity of responsibility in matters of policy. This seems to support Powell and Whitten (1993)’s findings further.

Discussion

Unfortunately, the results explained above do not appear to support either

Hypothesis 6.1 or Hypothesis 6.2. It seems that the Powell and Whitten’s (1993) work which shows that voters in federal systems can determine who is responsible for what in politics still holds true. While Ulbig and Miller (2012) make an interesting discovery in

American states, this reasoning may not be applicable to the Canadian case. This may be the case for two reasons.

First of all, the findings of Ulbig and Miller (2012) may be limited to the

American states and the cases of governors. This could be due to the fact that state and national party politics and provincial party politics may be seen as more closely linked than in this particular case of political scandal. When it was found that federal advertising dollars were making their way into Québec Liberal party coffers, this was seen as a serious problem. American party systems at the state level are more likely to look similarly to the federal level of politics than are Canadian systems which, as stated previously, differ significantly from their federal counterparts.

Canadians may also just find scandal to be an area which is highly visible and easier to keep separate than other policy areas or political events. Cutler (2004, 2008) identifies certain policies as being easily separate in the minds of Canadians. His studies show that Canadians have a difficult time determining who is responsible for social

125 policy in Canada but are easily able to separate provincial economic policies from federal economic policies.

Looking at Figures 6.1 and 6.2 it appears clear that in some ways, all Canadian provinces looked similarly in regards to reports of political efficacy and views of politicians. In Figure 6.1 it seems that in each province, the average respondent feels less politically efficacious in 2006 than in 2004 and most respondents, regardless of province, indicated a higher score of political efficacy in 2004 versus 2008. This seems consistent with the findings in Chapter Five about the Sponsorship Scandal.

The views about politicians, however, do not seem consistent however. Figure

6.2 indicates that respondents in all provinces viewed politicians more favorably in 2006 than 2004. Once more, the mean respondent from all provinces looks consistent across all of Canada and follows the same pattern of responses. This appears as a national rather than provincial trend. Still, it does seem odd that respondents were rate politicians more warmly in 2006 at the height of the Sponsorship Scandal.

The first reason for this may be that the actual impact of even great political scandals such as the Sponsorship Scandal may have specific, limited, and short-ranging implications. Perhaps by the time the 2006 surveys were actually conducted, people had moved on past the scandal and were focusing on new and more exciting political issues.

By this time, Stephen Harper and the Conservatives were promising change and were on the tails of winning the 2006 election. Excitement about a possible Conservative government may have become a source of hope and energy for voters who were tired of media coverage of the Sponsorship Scandal. Thus, while political efficacy may seem to

126 be affected people did not harbor bitter feelings towards all politicians when only some had been found to be involved in the Sponsorship Scandal.

Similarly, it could be that voters, in general, saw the Gomery Commission and ethics reforms, which occurred during this period of time to be evidence that the government was actually trying to improve transparency. As has been discussed in the literature review, government often acts in response to greater public demands for the ending of corruption and transparency and, in general, ethics reforms are met with great levels of public approval. As the Federal Accountability Act was passed in 2006 and was a major campaign promised made by the Conservative Party, it makes sense that perhaps people could think by 2006 their best interest was being taken into consideration by those hoping to lead in parliament.

While these could be likely reasons for these results, showing support for these hypotheses is beyond the span of this analysis and would require further study and testing. The implication of the aforementioned research, as this analysis currently asserts, is simply that there does not seem to be any support for the argument that scandal at the provincial level is able to translate to larger views on politics. In the above tests, there was no statistically significant support for the idea that people who provincially- experience a scandal feel less efficacious or less positively about politicians.

127

Figure 6.1

Mean Political Efficacy by Province and by Year 4

3.5

3

2.5

2

1.5

1

0.5

0 2004 2006 2008

AB BC QC MB NB NF NS ON PEI SK

128

Figure 6.2

Politicians Thermometer by Province & by Year 100

90 AB 80 BC 70 MB 60 NB 50 NF 40 NS 30 ON PEI 20

Mean Politicians Thermometer Value Thermometer Politicians Mean QC 10 SK 0 2004 2006 2008

129

Table 6.1: Descriptive Statistics N Mean Standard Minimum Maximum Deviation Politician 1595 9.207 21.886 -80 100 Thermometer 06- 04 Politician 1073 -3.598 19.934 -79 65 Thermometer 08- 06 Politician 1179 4.869 22.415 -76 85 Thermometer 08- 04 Political Efficacy 1925 -.190 0.989 -3 3 06-04 Political Efficacy 1185 .144 1.054 -3 3 08-0 Political Efficacy 1199 -.036 1.102 -3 3 08-04 Scandal Binary 9369 .619 .490 0 1 Age Squared 9503 2950.242 3.173 324 11236 News Attention 4018 3.431 0.816 0 10 2006 New Attention 3678 4.192 3.184 0 10 2008 Education 9518 1..900 0.816 1 3 Liberal- 3786 -1.508 40.304 -100 100 Conservative Thermometer 06- 04 Liberal- 1207 -6.287 39.784 -100 100 Conservative Thermometer 08- 06

130

Table 6.2: Scandal Binary

Province Year Scandal? AB 2004 NO BC 2004 YES MB 2004 YES NB 2004 YES NF 2004 NO NS 2004 NO ON 2004 YES PE 2004 NO QC 2004 YES SK 2004 YES AB 2006 YES BC 2006 YES MB 2006 YES NB 2006 NO NF 2006 YES NS 2006 YES ON 2006 NO PEI 2006 NO QC 2006 YES SK 2006 YES AB 2008 YES BC 2008 YES MB 2008 NO NB 2008 YES NF 2008 YES NS 2008 NO ON 2008 YES PEI 2008 NO QC 2008 YES SK 2008 YES

131

Table 6.3: Changes in Political Efficacy 2006-2004 2008-2006 2008-2004

Coefficien SE Coefficien SE Coefficien SE t t t Scandal .099** .049 -.067 .064 .111* .064 Binary Age Squared -.00002* .00001 .00002 .00002 -.00002 .0000 2 News .01 .007 .024** 0.01 .004 .01 Education .003 .029 .014 .038 .046 .040 Liberal- .001* .0006 -.007*** .0008 -.007*** .0008 Conservative Thermometer Constant -.221*** .086 -.047 0.12 -.195 .125 F 2.23** 20.47*** 15.20*** R-Squared .0061 .082 .062 N 1820 1147 1157 P-values indicated are *** significance at .01 or greater, **significance between .01 and .05, *significance between .05 and up to .10

132

Table 6.4: Changes in Politicians Thermometer 2006-2004 2008-2006 2008-2004

Coefficien SE Coefficien SE Coefficien SE t t t Scandal .234 1.165 2.02 1.3 2.338* 1.383 Binary Age Squared .0009*** .0003 -.001** .0003 .0001 .0004 News -.051 .179 .102 .2 -.180 .209 Education -1.876*** .685 1.122 .778 -.591 .826 Liberal- -.079*** .013 -.037** .016 -.082*** .017 Conservative Thermometer Constant 9.909*** 2.076 -5.018** 2.435 4.454 2.60 F 11.95*** 3.03*** 3.03*** R-Squared .038 .014 .014 N 1518 1044 1044 P-values indicated are *** significance at .01 or greater, **significance between .01 and .05, *significance between .05 and up to .10

133

Chapter Seven

Conclusion and Future Questions

Conclusions

The results of this study demonstrate that there is much to be gained from the study of federal and provincial scandals in Canada. The Sponsorship Scandal was an important political event in Canadian history that was able to change the direction of the long-governing Liberal Party. Moreover, the Sponsorship Scandal should be seen an important event for those both inside and outside Canada. Based on the results in

Chapter Five, it is clear that the Sponsorship Scandal had strong impacts on individual respondents. This study has extended the debate on the impact of this important political scandal by incorporating a quantitative approach to studying scandal and has attempted to link concerns about scandal at the federal and provincial level.

While data limitations may have hampered the robustness of the results between scandal and policy adoption, there appeared to be a relationship between Canadian provincial scandals and the passage of new or additional lobbying regulations in the

Canadian provinces. This supports previous findings in the American public policy literature. Moreover, these findings contribute empirically to a part of the comparative public policy literature that lacks empirical, quantitative studies that go beyond descriptive accounts of one particular political scandal. This study has shown that more needs to be done quantitatively to assess lobbying adoptions in other nations or cross- nationally. By widening the results across provinces, this hypothesis could be tested across time and cases in a way that had not yet been attempted. These results are not just limited to the findings about scandal, either. Instead, the results also support that regional

134 diffusion may impact policy in this area similarly to findings for policy innovations in the

United States.

The results from Chapter Five suggest, similarly that there is a relationship between attitudes regarding political scandal and salient changes in public opinion.

Respondents expressing more anger towards the Sponsorship Scandal were more likely to express feelings of lower political efficacy, more negative views of politicians, and less confidence in government. This further supports findings of studies done throughout

Europe and the U.S. There are important implications for the health of a democracy following a scandal. Political scandals can impact not only broad conceptions of democracy but also views of institutions. These events also can harm individual conceptions of the ability of citizens to wield power and can damage the views of crucial political institutions. Because a healthy democracy is deemed essential, these findings should concern political scientists.

Finally, the results of Chapter Six did not appear to indicate a relationship between provincial scandals and views of federal politics. These findings also did not support the assumption that those citizens in recently “scandalized” provinces would behave differently than citizens from provinces which had recently been free of scandal.

This supports the findings of many studies regarding federal and provincial accountability in Canada and other federal states. While this disputes the recent findings which suggest that evaluations of candidates and institutions of the federal level should be affected by state and local scandals in the American states, these results were not found in this analysis. Instead, it would appear that Canadians can determine who is responsible for scandal in Canada. This seems logically consistent with much of the

135 literature on federal accountability in Canada. This research indicates that in policy areas where responsibility is clear, Canadians can appropriately credit blame to the federal or provincial government.

Larger Implications

These findings are not only important because they showcase the results of a major event or major set of events in Canada. They are also novel because, as stated, they fill a void in the larger comparative politics literature. With these results, it is possible to add another quantitative, empirical study about policy adoption into a growing community of political scientists trying to expand policy adoption and public policy literature to other cases outside of the U.S. Finally, the Canadian case adds another layer to the puzzle, allowing for a greater study of federal cases. While not all the hypotheses regarding federalism were supported, the regional diffusion findings in Chapter Five suggest that more studies should be conducted on this implication.

Two other implications also should be outlined here as they have helped advance the study of scandal and other parts of comparative politics. The first is the development of a definition of scandal that is both inclusive and also feasibly adhered to which takes into account the importance of media but also the existing knowledge on types of political scandals. By excluding sex and other types of scandals, the definition has provided perhaps a better understanding of political scandal.

The second implication is that of data generation. The data used for Chapters

Four and Six will be made available to other scholars in the future. While the sample currently accounts for 20 years of time, the codebook and strategy could be updated by

136 another researcher to ensure scandals continue to be coded in the provinces. Also, the applications for this scandal variable are widespread. Perhaps future studies could focus on voting behavior and election results or the adoption of other types of legislation at the time of scandal. This is a contribution which should aid in future research in Canadian politics and comparative politics.

Future Questions

The research here is limited to the time and space allowed for this dissertation but that does not mean there are not still things to be learned. The study here is limited by data, as are most. It is a hope that as time goes by, more panel data can become available and the approach taken in Chapter Six can be expanded to cover more time points.

Perhaps the 2011 CES or upcoming 2015 CES could allow for more data points.

Likewise, the data from Chapter Four could be expanded as time progresses to widen the sample and perhaps produce more robust results. The data limitations restricted some of these models significantly but that does not mean in the future that they should not be tested again. It is possible that a stronger result could be identified if the sample was expanded.

Likewise, this study does focus primarily on one case and a set of cases within. A truly cross-national study could be expanded with more time and resources to apply the same scandal definition to other cases. While it may not be helpful to include developed democracies and developing democracies in the same sample, it may be useful to research both the adoption of lobbying policies in developed democracies and developing democracies in the future. Because more time should be spent on understanding the

137 reasons for policy adoption of ethics reforms and the types of reforms adopted, future research needs to expand to include more and diverse cases.

Another future research question concerns the impact of lobbying and ethics reforms. While this question was beyond the bounds of this dissertation, it is an important one for future research. Countries around the world have now adopted regulations but cross-national studies have not been conducted similarly to the ones conducted in the United States. It is imperative that more research be conducted to determine whether these regulations help or hurt democracies around the world. This is especially true in the developing world where such regulations may significantly reduce corruption.

While much has been explored regarding lobbying and ethics commissions in the

United States, lessons about Canada can help expand knowledge of federalism, policy diffusion, and the importance of scandal in policy adoption. American and Canadian federalism are not identical but it is worth noting many similarities do exist between these two nations. Policy diffusion is an area of research that has taken off quite dramatically in American politics but quantitative studies of policy diffusion are still relatively new in

Canada. Studies of Canadian policy diffusion may not only teach Canadians but also

Americans. Also, similar strategies of data collection have used news articles to code for scandals at the state level, but few analyses have had as expansive a sample as the one seen here. Very little research on lobbying policy adoption has included a 20 year time trend. Even fewer include all 50 states. Such a sample could speak loudly about recent lobbying policy changes in the American states. If time and money were no object, this analysis would have included another chapter devoted to lobbying policy adoptions in the

138

American states. After all, this has been an important issue recently in states like

Missouri. I believe this would have been a very salient addition to this research.

A better understanding of lobbying regulations, why they are passed, and how they are able to help may have real, practical effects in countries throughout the world.

Knowledge may empower citizens to act on their own best interests and to pressure governments to enact stringent regulations to protect the citizenry from an oligarchy of special interests. Only more evidence can suggest this, however. In a similar vein, the research suggests that, at the very least, scandal can be harmful to democracy in a significant way. Thus, if these regulations do work, they should be implemented to protect democracy and ensure that citizens are being represented in the best way by their elected officials and civil servants.

139

Appendix One: Chapter Four Cox Models Table 7.1: Cox Model with Scandal Binary Variable Variable Hazard Robust P Value 95% Ratio Standard Confidence Error Interval Scandal Binary 1.285 0.479 0.501 (.620, 2.668) Economy 1.000 0.000008 0.026 (.99997, .999998) Region 0.554 0.467 0.484 (.106, 2.900) Population 1.000 0.00000039 0.021 (1, 1.000002) Chi Square 13.23 0.01 N 200

Table 7.2: Cox Model with Scandal Percentage Variable Variable Hazard Robust P Value 95% Ratio Standard Confidence Error Interval Scandal Percentage 1.194 0.515 0.682 (.512, 2.781) Economy 1.000 0.000009 0.02 (.99996, .999998) Region 0.612 0.484 0.535 (.129, 2.84) Population 1.000 0.0000004 0.019 (1, 1.000002) Chi Square 15.72 0.0034 N 200

140

Appendix Two: CES Questions Used Questions for the 2006 CES (Chapter Five) VARIABLE QUESTION RESPONSES Efficacy I don't think the 4=Strongly Disagree, government cares much 3=Somewhat Agree, about what people like me 2=Somewhat Disagree, 1= think. Strongly Disagree; DK/other coded as missing Politicians And on the same scale, 0=strongly dislike, Thermometer how do you feel about 100=strongly dislike; POLITICIANS IN DK/other coded as missing GENERAL? Confidence in How much confidence do 4=a great deal, 3= some, Government you have… in the federal 2=very little 1=none at all; government? DK/other coded as missing Anger About Now some questions 1=not very angry at all, Sponsorship Scandal about the Sponsorship 2=not very angry, Scandal. Does it make 3=somewhat angry, 4=very you very angry, somewhat angry; DK/other coded as angry, not very angry, or missing not angry at all? Political Volunteer Has the respondent ever 1= Yes, 0=all else volunteered for a political party? Community Volunteer Has the respondent ever 1= Yes, 0=all else volunteered for non-profit or community organizations? Age Year of birth Coded by subtracting age of birth from 2008 and then squaring. News Attention How much attention have 0 to 10 scale where 0= no you paid to news about attention and 10= a great the federal election in a deal of attention newspaper?

141

Education What is the highest level Recoded as 1=high school of education you have education or less, 2=some completed? post-secondary ed, 3=a bachelor's and above Liberal-Conservative How do you feel about the -100 to 100 scale; 0= really Thermometer federal Liberal party? dislike the party to 100= How do you feel about the really like the party federal Conservative

party. Score of Conservative Party subtracted from Liberal Party

Questions for the Panel Study (Chapter Six) VARIABLE QUESTION RESPONSES

Efficacy I don't think the government 4=Strongly Disagree, cares much about what 3=Somewhat Agree, people like me think. 2=Somewhat Disagree, 1= Strongly Disagree; DK/other coded as missing Politicians Politicians are ready to lie to 1=Strongly Disagree, Ready to Lie get elected. 2=Somewhat Agree, 3=Somewhat Disagree, 4= Strongly Disagree;DK/other coded as missing Politicians 0-100 Scale 0=strongly dislike, 100=strongly Thermometer slike; DK/other coded as missing Political Has the respondent ever 1= Yes, 0=all else Volunteer volunteered for a political party? Community Has the respondent ever 1= Yes, 0=all else Volunteer volunteered for non-profit or community organizations? Age Squared Year of birth Coded by subtracting age of birth from 2008 and then squaring.

142

News Attention How much attention have 0 to 10 scale where 0= no you paid to news about the attention and 10= a great deal of federal election in a attention newspaper? Education What is the highest level of Recoded as 1=high school education you have education or less, 2=some post- completed? secondary ed, 3=a bachelor's and above Liberal – How do you feel about the -100 to 100 scale; 0= really Conservative federal Liberal party? How dislike the party to 100= really Thermometer do you feel about the federal like the party Conservative party.

Score of Conservative Party subtracted from Liberal Party

143

Appendix Three: Political Scandal Codebook

A. Date – Month/Day/Year

B. Province

C. Sources

D. Coding Scandal Dichotomy 1 scandal coded 0 scandal not coded

E. Rank in file legislators involved– were they involved 0=no, 1= yes. Dichotomy if coded

F. Number of Cabinet Members or Leadership position Coded - number involved 0+ for each coded

G. Premier Involved, Opposition Party Leader involved – was the premier involved or opposition party leader involved? 0=no, 1= yes. Dichotomy if coded

H. Civil Service Members Involved – were civil servants involved 1=yes, 0=no? Dichotomy if coded

I. Explanation Type 1 Episodic 2 Thematic

J. Type of Article 1 News Feature 2 Editorial/Op-Ed 3 News Brief 4 Miscellaneous 5 Letters to the Editor 6 Not Coded

M National Scandal or other Scandal 1 if yes, 0 if no

144

Bibliography

Ainsworth, Scott H. 1997. “The Role of Legislators in the Determination of Interest

Group Influence.” Legislative Studies Quarterly 22(4): 517-533.

Allen, Nicholas. 2010. “Keeping MP's Honest? Ethics Reforms in the British House of

Commons.” Public Integrity 12(2): 105-123.

Almond, Gabriel, and Sidney Verba. 1963. The Civic Culture. Princeton, NJ: Princeton

University Press.

Anonymous. 2004. “Public must weigh all facts.” 8 April 2004. Telegraph-Journal.

Antecol, Michael, and James W. Endersby. 1999. “Newspaper Consumption and Beliefs

about Canada and Quebec.” Political Communication 1(1): 95-112.

Alt, James E., and Robert C. Lowry. 1994. “Divided Government, Fiscal Institutions and

Budget Deficits: Evidence from the States.” American Political Science Review

88:811-28.

Atkeson, Lonna Rae, and Randall W. Partin 1995. “Economic and Referendum Voting: A

Comparison of Gubernatorial and Senatorial Elections.” American Political

Science Review 89(1): 99-107.

Balch, George I. 1972. “Multiple Indicators in Survey Research: The Concept ‘Sense of

Political Efficacy’.” Political Methodology 1(2): 1-43

Bakvis, Herman, and Grace Skogstad. 2002. Canadian Federalism: Performance,

Effectiveness and Legitimacy. Don Mills, Ont.: Oxford University Press.

145

Barrilleaux, Charles, and Ethan Bernick. 2002. “Deservingness, Discretion, and the State

Politics of Welfare Spending, 1990-96.” State Politics and Policy Quarterly 3:1-

22.

Benndesen, Morten, and Sven E. Feldmann. 2006. “Information Lobbying and Political

Contributions.” Journal of Public Economics 9(4): 631-656.

Bennett, Leslie E. 1999. “One Lesson From History: Appointment of Special Counsel

and the Investigation of the Teapot Dome Scandal”. Washington, D.C.: Brookings

Institution.

Benoit, Kenneth and Michael Laver. 2006. Party Policy in Modern Democracies.

London: Routledge.

Berry, Francis Stokes, and William D. Berry. 1990. “State Lottery Adoptions as Policy

Innovations: An Event History Analysis.” American Political Science Review

84(2): 395-415.

Bhatti, Yosef, Kasper M. Hansen, and Asmus Leth Olsen. 2013. “Political Hypocrisy:

The Effect of Political Scandals on Candidate Evaluations.” Acta Politica 48: 408

428.

Blais, Andre. 2000. To Vote or Not to Vote. Pittsburgh: Univ. of Pittsburgh.

Blais, Andre. 2006. “What Affects Voter Turnout?” Annual Review of Political Science

9: 111 125.

Black, Peter. 2004. “Bloc stands to benefit if voters chose to punish Liberals.” 21 May

2004. Telegraph-Journal.

146

Boatright, Robert. 2009. “Interest Group Adaptations to Campaign Finance Reform in

Canada and the United States.” Canadian Journal of Political Science 42(1):17

43.

Bowler, Shaun, and Jeffery A. Karp. 2004. “Politicians, Scandals, and Trust in

Government.” Political Behavior 26(3): 271-287.

Bowler, Shaun, and Todd Donovan. 2013. “Civic duty and turnout in the UK referendum

on AV: What shapes the duty to vote?” Electoral Studies 32(2): 265-273.

Brinig, Margaret F., Randall G. Holocombe, and Linda Schwartzstein. 1993. “The

Regulation of Lobbyists.” Public Choice 77: 377-384.

Bullock, Joyce. 1994. “State Lobby Laws in the 1990s.” In The Book of the States, vol.

30. Lexington, KY: Council of State Governments.

Butt, Sarah, and John Curtice. 2010. “Duty in Decline? Trends in Attitudes in Voting.” In

British Social Attitudes: The 26th Report, eds. Alison Park, Elizabeth Clery,

Miranda Phillips, Katarina Thomson, Sarah Butt, and John Curtice. Thousand

Oaks, CA: Sage.

Canada, Government of. 2016. Statistics Canada. http://www.statcan.gc.ca/start-debut

eng.html.

Canadian Legal Information Institute. 2016. https://www.canlii.org/en/.

Callier, James. 2010. “Citizen Trust, Political Corruption, and Voting Behavior:

Connecting the Dots.” Politics and Policy 38(5): 1015-1035.

Campbell, Angus, Phillip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960.

The American Voter. New York: Wiley.

147

Carroll, Barbara Wake and Ruth J. E. Jones. 2000. “The Road to Innovation,

Convergence or Inertia: Devolution in Housing Policy in Canada.” Canadian

Public Policy 26(3): 277-293.

Carsey, Thomas M., and Gerald C. Wright 1998. “State and National Factors in

Gubernatorial and Senatorial Election: a Rejoinder.” American Journal of

Political Science 42(3): 994 1002.

Chanley, Virginia A., Thomas J. Rudolph, and Wendy M. Rahn. 2000. “The Origins and

Consequences of Public Trust in Government: A Time Series Analysis.” Public

Opinion Quarterly 64: 239-256.

Chari, Raj, John Hogan, and Gary Murphy. 2007. Regulating Lobbyists: A Comparative

Analysis of the United States, Canada, Germany, and the European Union. The

Political Quarterly 78:422-438.

Chari, Raj, John Hogan, and Gary Murphy. 2010. Regulating Lobbying: A Global

Comparison: Manchester University Press.

Chari, Raj, and Sylvia Kritzinger. 2006. Understanding EU Policy Making: National

Versus European Sovereignty: Pluto Press.

Chubb, John E. 1988. “Institutions, the Economy, and the Dynamics of State Elections.”

American Political Science Review 82(1): 133-145.

Citrin, Jack, and Christopher Muste. 1999. “Trust in Government.” In Measures of

Political Attitudes, ed. John P. Robinson, Phillip R. Shaver, and Lawrence S.

Wrightman. San Diego: Academic Press. 465-532.

148

Coen, David. 1998. “The European Business Interest and the Nation State: Large-Firm

Lobbying in the European Union and Member States.” Journal of Public Policy

18(1): 75-100.

Coen, David. 2007. “Empirical and Theoretical Studies in EU Lobbying.” Journal of

European Public Policy 14(3): 333-345.

Cohen, Jeffery E. 1983. “Gubernatorial Popularity in Nine States.” American Politics

Quarterly 11: 219-235.

Cohen, Jeffery E., and John D. King. 2004. “Relative Unemployment and Gubernatorial

Popularity.” Journal of Politics 66(4): 1267-1282.

Commission, Federal Election. 2012. “Federal Contribution Limits 2012.”

Commission, Federal Election. 2015. “ The Federal Election Campaign Laws: A Short

History.” Federal Election Commission http://www.fec.gov/info/appfour.htm.

Commission of Inquiry into the Sponsorship Program and Advertising Activities. 2006.

Restoring Accountability: Recommendations. Ottawa: Government of Canada.

Commission of Inquiry into the Sponsorship Program and Advertising Activities. 2006.

Who Is Responsible? Fact Finding Report. Ottawa: Government of Canada.

Commissioner of Lobbying of Canada. December 2014. “Survey of Lobbying Legislation

in Canada.” Office of the Commissioner of Lobbying.

Common Cause. 1991. “Common Cause issues poll.” Common Cause.

Converse, Phillip.E., 1972. “Change in the American electorate.” In The human meaning

of social change, eds. Phillip E. Converse and Angus Campbell.263-337.

Craig, Stephen C., Richard G. Niemi, and Glenn E. Silver. 1990. “Political Efficacy and

Trust: A Report on the NES Pilot Study Items.” Political Behavior 12:289-314.

149

Curtice, John and Roger Jowell 1997. “Trust in the Political System.” In. British Social

Attitudes eds. Roger Jowell. Farnham, UK: London.

Cutler, Fred. 2004. “Government Responsibility and Accountability in Federations.”

Publius 34(2): 19-38.

Cutler, Fred. 2008. “Whodunnit? Voters and Responsibility in Canadian Federalism.”

Canadian Journal of Political Science 41(3): 627-654.

Dalton, Russell J. 2002. Citizen Politics: Public Opinion and Political Parties in

Advanced Industrial Democracies. New York: Chatham.

Dalton, Russel J. 2008. The Good Citizen: How a Younger Generation is Reshaping

American Politics. Washington, DC: CQ Press.

Dancey, Logan. 2012. “The Consequences of Political Cynicism: How Cynicism Shapes

Citizens' Reactions to Political Scandals.” Political Behavior 34(3): 411-423.

Downs, Williams M. 1999. “Accountability Payoffs in Federal Systems? Competing

Logics and Evidence from Europe's New Federalism.” Publius 29(1): 87-110.

Duquette, Jerold J. 2002. “Campaign finance reform in the Bay State: Is cleanliness really

next to godliness.” In D. Schultz ed. Money, Politics and Campaign Finance

Reform in the States. Durham, NC: Carolina Academic Press.

Dur, Andreas. 2008a. “Interest Groups in the European Union: How Powerful Are

They?” West European Politics 31(6): 1212-1130.

Dur, Andreas. 2008b. “Measuring Interest Group Influence in the EU: A Note on

Methodology.” European Union Politics 9(4): 559-576.

Elazar, Daniel. J. 1984. American federalism: A view from the states, 3rd ed. New York,

NY: Harper and Row.

150

Ensign, David. 1997. “Reforming Public Integrity Laws in an Era of Declining Trust.” In

The Book of the States, vol. 31. Lexington, KY: Council of State Governments.

Franks, C.E.S. 1987. The Parliament of Canada. Toronto: University of Toronto.

Gélaneau, Francois and Karen L. Remmer. 2005. “Political Decentralization and

Electoral Accountability: The Argentine Experience 1983-2001.” British Journal

of Political Science 36(1): 133-157.

Gidengil, Elisabeth, Joanna Everitt, Patrick Fournier, Neil Nevitte. 2008. The 2004 2008

Canadian Election Study. http://www.queensu.ca/cora/ces.html.

Government of Canada. 2 February 2015. “The Lobbying Act.” Office of the

Commissioner of Lobbying https://lobbycanada.gc.ca/eic/site/012.nsf/eng/h_

00008.html#key_events.

Greene, Ian. 2006. “Review Issue: Commission of Inquiry into the Sponsorship Program

and Advertising Activities, Phase I Report and Phase II Report.” Canadian Public

Administration 49(2): 220-232.

Gray, Virginia and David Lowery. 1993. “The Density of State Interest Group Systems.”

The Journal of Politics 55(1): 191–206.

Gray, Virginia and David Lowery. 1995. “How Some Rules Just Don’t Matter: The

Regulation of Lobbyists.” Public Choice 91(2): 139–47.

Gray, Virginia and David Lowery. 1999. “Interest Representation in the States.” In

American State and Local Politics: Directions for the 21st Century, eds. Ronald

E. Weber and Paul Brace. New York: Chatham House.

Hall, Richard L. and Alan A. Deardorff. 2006. “Lobbying as a Legislative Subsidy.”

American Political Science Review 100(1): 69-84.

151

Harsanyi, Fruzsina M., and Susan Schmidt. 2012. “Creating a Public Affairs Function in

Countries Without a Public Affairs Culture.” Journal of Public Affairs 12 (1) 86–

97.

Hibbing, John R. and David Marsh. 1987. “Accounting for the Voting Patterns of British

MPs on Free Votes.” Legislative Studies Quarterly 12(2): 275-294.

Hoogenboom, Ari. 1978. “Did Gilded Age Scandals Bring Reform?” In Before

Watergate, eds. Abraham S. Eisenstadt, Ari Hoogenboom, and Hans L. Trefousse.

New York: Brooklyn College Press.

Integrity Commissioner, Office of the. 2013. “Members’ Integrity.” Office of the Integrity

Commissioner http://www.oico.on.ca/home/mpp-integrity.

Jasiecki, Kryzsztof. 2006. “Regulating lobbying in Poland: Background, scope and

expectations.” Working Paper. http://www.coe.int/T/DG1/LegalCooperation/

Economic crime/cybercrime/cy%20activity%20Interface2006/143%20_2006_-if

rep%20jasie.pdf.

Jordan, Grant. 1998. “Towards Regulation in the UK: From 'General Good Sense' to

'Formailsed Rules'.” Parliamentary Affairs 51: 524-537.

Jordan, Andrew, Wurzel, Rudiger K. and Zito, Anthony R., 2003. Has governance

eclipsed government? Patterns if environmental instrument selection and use in

eight states and the EU (No. 03-15). CSERGE Working Paper EDM.

Julliet, Luc. 2004. The Office of Ethics Commissioner, Accountability, and Public Trust.

In G. B. Doem ed. How Ottawa Spends, 2004-2005: Mandate Change and

Continuity in the Paul Martin Era. 296–316. Toronto: McGill-Queen's University

Press.

152

Kaufield, Kathy. 29 June 2004. “Liberal cabinet minister holds his riding easily.”

Telegraph-Journal.

Kaufield, Kathy. 19 October 2004. “Liberal wants investigation into MLA's fundraising

call; 'There's no intention whatsoever of saying, come to my dinner or you don't

get a grant . . . ,' MLA responds.” Telegraph-Journal.

Keane, Bernard. 3 April 2008. “Why Faulkner's Lobbyist Registry Won't Work.” Crikey.

http://www.crikey.com.au/2008/04/03/why-faulkners-lobbyist-register-wont

work/.

Kiousis, Spiro. 2003. “Job Approval and Favorability: The Impact of Media Attention to

the Monica Lewinsky Scandal on Public Opinion of President Bill Clinton.” Mass

Communication and Society 6(4): 435-451.

Krieg, Gregory. 7 January 2016. “Missouri lawmaker – Sex with lobbyists is a ‘gift’.”

Fox 2 St. Louis http://fox2now.com/2016/01/07/missouri-lawmaker-sex-with

lobbyists-is-a-gift/.

Kumlin, Staffan and Peter Essaisson. 2011. “Scandal Fatigue? Scandal Elections and

Satisfaction with Democracy in Western Europe, 1977–2007.” British Journal of

Political Science 42: 263-282.

Lane, Robert E. 1962. Political Ideology. New York: Free Press.

Lee, Francis L. F. 2014. “How Citizens React to Political Scandals Surrounding

Government Leaders: A Survey Study in Hong Kong.” Asian Journal of Political

Science 22(1): 1-19.

153

Legislative Assembly of Ontario. 2015. Lobbyists Registration Act of 1998.

https://www.ontario.ca/laws/statute/98l27?search=Lobbyists+Registration

+Act#BK1.

Leiter, Debra and Michael Clark. 2015. “Valence and satisfaction with democracy: A

cross national analysis of nine Western European democracies.” European

Journal of Political Research 54(3): 543-562.

Lewis-Beck, Michael. 1988. Economics and Elections: The Major Western Democracies.

Ann Arbor, MI: University of Michigan Press.

Longley, Neil. 1999. “Voting on Abortion in the House of Commons: A Test of

Legislator Shirking.” Canadian Public Policy 25(4):503—21.

Lowi, Theodore. 1988. “Forward.” In The Politics of Scandal: Power and Process in

Liberal Democracies, eds. Andrei S. Markovits and Mark Siverstein. New York:

Holmes and Meier.

McCleery, Shawn M. 2007. “Not My Fault: Ministerial Responsibility and the

Sponsorship Scandal.” Master’s Thesis. Carleton University: Ottawa, ON.

Miller, Arthur H. 1999. “Sex, Politics, and Public Opinion: What Political Scientists

Really Learned from the Clinton-Lewinsky Scandal.” PS: Political Science and

Politics 32(4): 721-729.

Niemi, Richard, Stephen Craig and Franco Mattei. 1991. “Measuring Internal Efficacy in

the 1988 National Election Study.” American Political Science Review. 85(4):

1407-1413.

Newmark, A. J. 2005. “Measuring State Legislative Lobbying Regulation, 1990-2003.”

State Politics and Policy Quarterly 5(2): 182-191.

154

Newton, Kenneth. 2006. “Political Support: Social Capital, Civil Society, and Political

and Economic Performance.” Political Studies 54(4): 846-864.

Opheim, Cynthia. 1991. “Explaining the Differences in State Lobby Regulation.” The

Western Political Quarterly 44, no. 2: 405–421.

Overby, L. Marvin, Chris Raymond and Zeynep Taydas. 2011. “Free Votesm MPs, and

Constituents: The Case of Same-Sex Marriage in Canada.” American Review of

Canadian Studies 41(4): 465-478.

Overby, L. Marvin, Raymond Tatalovich and Donley T. Studlar. 1998. “Party and Free

Vote in Canada: Abortion in the House of Commons.” Party Politics 4(3):381

392.

Ozymy, Joshua. 2010. “Assessing the Impact of Legislative Lobbying Regulations on

Interest Group Influence in U.S. State Legislatures.” State Politics and Policy

Quarterly 10(4): 397–420.

Ozymy, Joshua. 2013. “Keepin’ on the Sunny Side: Scandals, Organized Interests, and

the Passage of Legislative Lobbying Laws in the American States.” American

Political Research 41(1): 3–23.

Pattie, Charles, and Ron Johnson. 2001. “Losing the voters' trust: Evaluations of the

political system and voting at the 1997 British General Election.” British Journal

of Political Science 3(2): 191-222.

Peters, John G., and Welch, Susan. 1980. “The Effects of Charges of Corruption on

Voting Behavior in Congressional Elections.” American Political Science Review

74(3): 697-708.

155

Pomey, Marie-Pascale, et al. 2010.”Research article Does accreditation stimulate change?

A study of the impact of the accreditation process on Canadian healthcare

organizations.” Implementation Sci 31(5): 1-14.

Powell, G. Bingham and Guy D. Whitten. 1993. “A Cross-National Analysis of

Economic Voting: Taking Account of the Political Context.” American Journal of

Political Science 37(2): 391-414.

Praino, Rodrigo, Daniel Stockemer and Vincent G. Moscardelli. 2013. “The Lingering

Effect of Scandals in Congressional Elections: Incumbents, Challengers, and

Voters.” Social Science Quarterly 94(4): 1045-1061.

Primo, David M. and Jeffrey Milyo. 2006. “Campaign Finance Laws and Political

Efficacy: Evidence from the States.” Election Law Journal: Rules, Politics, and

Policy. 5(1): 23-39.

Putnam, Robert D. 1995. “Tuning In, Tuning Out: The Strange Disappearance of Social

Capital in America.” PS: Political Science and Politics 28(4): 664-683.

Putnam, Robert D. 2000. Bowling Alone: The Collapse and Revival of the American

Community. New York: Simon and Schuster.

Rabe, Barry G. 2 June 2005. Moral Super Power or Policy Laggard? Translating Kyoto

Protocol Ratification into Federal and Provincial Climate Policy in Canada.

Paper presented at the Annual Meeting of the Canadian Political Science

Association in London, Ontario.

Rechtman, Rene. 1998. ”Regulation of Lobbyists in Scandinavia: A Danish

Perspective.” Parliamentary Affairs 51 (4): 579‐586.

156

Rosenson, Beth A. 2003. “Against Their Apparent Self-Interest: The Authorization of

Independent State Legislative Ethics Commissions, 1973-96.” State Politics and

Policy Quarterly 3(1): 42-65.

Rosenson, Beth A. 2006. “The Impact of Ethics Laws on Legislative Recruitment and the

Occupational Composition of State Legislatures.” Political Research Quarterly

59(4): 619-627.

Ruderman, Nick and Neil Nevitte. 2016. “Assessing the Impact of Political Scandals on

Attitudes toward Democracy: Evidence from Canada’s Sponsorship Scandal.”

Canadian Journal of Political Science 49(1): 1-20.

Rush, Tony. 1998. “The Canadian experience: the Lobbyists Registration Act.”

Parliamentary Affairs 51: 516-523.

Rosenthal, Alan. 2000. The Third House: Lobbyists and Lobbying in the States, 2nd ed.

Washington, DC: CQ Press.

Sarmiento-Mirwaldt, Katja, Nicholas Allen, and Sarah Birch. 2014. “No Sex Scandals

Please, We’re French: French Attitudes towards Politicians’ Public and Private

Conduct.” West European Politics 37(5): 867-885.

Schattschneider, E.E. 1960. The Semisovereign People: A Realist's View of Democracy in

America. Holt, Reinhart, and Winston: New York.

Shah, Dhavan V., Mark D. Watts, David Domke, and David P. Fan. 2002. “News

Framing and Cueing of Issues Regimes: Explaining Clinton's Public Approval in

Spite of Scandal.” Public Opinion 66: 339-370.

157

Shea, Daniel M. 1999. “All Scandal Politics is Local: Ethical Lapses, the Media, and

Congressional Elections.” Harvard International Journal of Press/Politics 4(2):

45-62.

Smyth, David J and James Holmes. 2003. “Presidential Popularity: What Matters Most,

Macroeconomics, or Scandals.” Applied Economics Letters 10L 585-88.

Stark, David., and Balazs Vedres. 2012. “Political Holes in the Economy: The Business

Network of Partisan Firms in Hungary.” American Sociological Review 77(5):

700-722.

Strickland, James. 2014. “The Semisovereign Lobbyists: Interest Regulations and Group

Density in the American States, 1990 – 2012.” Presented at the State Politics and

Policy Conference, Indiana University Bloomington, Bloomington.

http://sppc2014.indiana.edu/Papers/Strickland.pdf.

Thomas, Clive S. 1998. “Interest Group Regulation across the United States: Rationale,

Development and Consequences.” Parliamentary Affairs 51(4): 500–515.

Thompson, John B. 2000. Political Scandal. Cambridge: Blackwell.

Ulbig, Stacy G. and Nancy M. Miller. 2012. “The Coingate Effect: The Impact of

Scandal on Attitudes toward State and Federal Political Actors.” Social Science

Journal 49(1): 61-71.

Veksler, Albert. (2011). “Opaque Transparency: An Analysis of the Israeli Lobbying

Regime of 2008.” Journal of Public Affairs 12(4): 270-278.

Villoria, Manuel, Gregg G. Van Ryzin and Cecilia F. Lavena. “Social and Political

Consequences of Administrative Corruption: A Study of Public Perceptions in

Spain.” Public Administration Review 73(1): 85-94.

158

Wagner, Marcus, Jessica Tarlov and Nick Vivyan. 2014. “Partisan Bias in Opinion

Formation on Episodes of Political Controversy from Great Britain.” Political

Studies, 62(1), 136-158.

Warhurst, John. (1998). “Lobbying the Target: Regulating Lobbying in Australia.”

Parliamentary Affairs 51(4): 538-550.

Welch, Susan, and Hibbing, John R. 1997. “The Effects of Charges of Corruption on

Voting Behavior in Congressional Elections, 1982-1990.” Journal of Politics

59(1): 226-239.

Wesselius, Erik. 2005. “High Time to Regulate EU Lobbying.” Consumer Policy Review

15(1): 13-18.

Witko, Christopher. 2007. “Explaining Increases in the stringency of State Campaign

Finance Regulation,1993-2002.” State Politics and Policy Quarterly 7(4), 369

393.

Yishai, Yael. 1998. “Regulation of Lobbying Groups in Israel.” Parliamentary Affairs,

51(4), 568-578.

159

Vita

Gwendolyn E. Sams became interested in politics and public policy before she could walk. She learned about Canadian politics while an undergraduate at Indiana

University, where she received her B.A. in Political Science with Highest and

Departmental Honors. After receiving her Bachelor’s, she enrolled at the University of

Missouri to study Political Science and Public Policy. While she was interested in many areas of public policy, she developed an even greater appreciation for Canadian public policy and politics as she worked with the Canadian Studies Center. The opportunities afforded by the CSC and the Kinder Forum allowed her to go to Canada multiple times to meet with academics, present research, and conduct studies there.

Gwendolyn is now a Research Analyst for the State of Missouri’s Department of

Corrections and responds to daily research requests on a variety of topics and helps write state legislative briefs. She resides in Columbia with her husband, who tolerates her love of Hoosier Basketball, the Cubs, and playoffs hockey, and her two dogs, Willis and

Chessie. She intends to remain dedicated to improving the lives of citizens through her public service for years to come and to ensuring the good people of Missouri will learn the virtues of poutine one day.

160